AI-Native Rebuild Opportunities: Failed/Inactive YC Startups¶
Prioritization Analysis | March 2026¶
Methodology¶
Each startup is scored on five dimensions:
| Dimension | Weight | What It Measures |
|---|---|---|
| AI Leverage | 25% | How much modern AI (LLMs, agents, voice, vision) reduces team size and improves the product vs. what was possible when the startup operated |
| Market Timing | 20% | Whether demand has grown since the original failure, and if the market is ready NOW |
| Small Team Fit | 20% | Whether 5 people can realistically build, ship, and scale this |
| Revenue Clarity | 20% | How clear and proven the path to revenue is |
| Defensibility | 15% | Ability to build moats through data, network effects, or vertical expertise |
Weighted Total = (AI Leverage x 0.25) + (Market Timing x 0.20) + (Small Team Fit x 0.20) + (Revenue Clarity x 0.20) + (Defensibility x 0.15)
Detailed Scoring: All 38 Startups¶
TIER: Agentic AI / Voice AI Opportunities¶
1. Opkit (S21) -- Healthcare AI Phone Call Automation¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 10 | Voice AI has undergone a revolution since 2021. Real-time speech-to-speech models, sub-200ms latency, and natural-sounding voices make healthcare phone call automation genuinely viable now. The entire product thesis is validated by the current generation of AI. |
| Market Timing | 9 | Healthcare admin costs are exploding. Staff shortages in medical offices are acute. Insurance verification, prior auth, and appointment scheduling calls are universally hated. CMS regulatory changes are pushing automation. Buyers are actively seeking solutions. |
| Small Team Fit | 8 | Voice AI infrastructure (ElevenLabs, Deepgram, OpenAI Realtime API) means the team builds orchestration and healthcare-specific logic, not foundational voice tech. HIPAA compliance adds overhead but is manageable. |
| Revenue Clarity | 9 | Per-call or per-minute pricing with clear ROI: a medical office spends $15-25/hour on staff making insurance calls. AI does it for pennies. B2B SaaS with usage-based pricing. Obvious buyer (practice managers, billing companies). |
| Defensibility | 7 | Healthcare-specific call flows, payer-specific knowledge graphs, and accumulated success rate data create a compounding advantage. Regulatory compliance is a barrier to entry. But voice AI commoditization is a risk. |
Weighted Total: 8.75
2. Struct (W23) -- Multi-lingual AI Voice Agents¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | Multi-lingual voice AI is now dramatically better. Real-time translation, code-switching, and accent handling have improved enormously. LLMs handle context and intent across languages well. |
| Market Timing | 7 | Global BPO market is massive and ripe for disruption. Multi-lingual support is expensive. However, the market is competitive with well-funded players (Bland AI, Vapi, etc.) already operating. |
| Small Team Fit | 7 | Voice infrastructure is commoditized. Multi-lingual adds complexity in testing and quality assurance across languages. Cultural nuance requires domain expertise per market. |
| Revenue Clarity | 8 | Clear B2B pricing: per-minute or per-call, displacing expensive multi-lingual call center agents ($8-15/hour in nearshore, more onshore). Enterprise contracts are large. |
| Defensibility | 5 | Multi-lingual is a feature, not a moat. Major voice AI platforms are adding language support rapidly. Without deep vertical focus, this becomes a commodity. |
Weighted Total: 7.35
3. Argovox (S22) -- Voice AI Agents for Patient Billing¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | Patient billing calls are highly structured and repetitive -- ideal for voice AI agents. LLMs handle the conversational flexibility needed for payment plan negotiations, balance inquiries, and insurance questions. |
| Market Timing | 9 | Patient bad debt is a top-3 concern for health systems. The No Surprises Act created complexity. Patient financial responsibility is growing. Revenue cycle management companies are desperate for automation. |
| Small Team Fit | 8 | Narrow scope (billing calls only) means less surface area. Can leverage existing voice AI infrastructure. HIPAA and PCI compliance add overhead but are well-understood. |
| Revenue Clarity | 9 | Revenue cycle management has proven willingness to pay. Per-call pricing with clear ROI: collections improve, staff costs drop. Can price as percentage of collections improvement. |
| Defensibility | 7 | Healthcare billing is deeply complex (payer rules, state regulations, facility-specific policies). Accumulated billing logic and success patterns create real defensibility. Switching costs are high once integrated with practice management systems. |
Weighted Total: 8.50
4. VOIQ (S15) -- Conversational AI Voicebot for Enterprise¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | Enterprise voice automation was impossible in 2015 -- IVR systems were terrible. Modern voice AI makes natural enterprise conversations feasible. However, "enterprise voicebot" is broad and unfocused. |
| Market Timing | 6 | The enterprise voice AI market exists but is crowded. Established players (Five9, Genesys, NICE) have added AI. Pure-play AI voice companies (Bland, Retell, Vapi) are well-funded. Timing was better 18 months ago. |
| Small Team Fit | 5 | "Enterprise" means long sales cycles, custom integrations, security reviews, SOC2, and dedicated support. This is hard with 5 people. |
| Revenue Clarity | 7 | Enterprise contracts are large but slow to close. Usage-based pricing works. But the broad positioning makes it hard to nail initial ICP. |
| Defensibility | 4 | Generic enterprise voice AI has no moat. Too many well-funded competitors. Without vertical focus, this is a losing position. |
Weighted Total: 6.20
5. Call9 (S15) -- Telehealth Emergency Medicine¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 6 | AI can assist with triage, documentation, and decision support, but emergency medicine requires licensed physicians. AI augments but cannot replace the core service. Regulatory constraints limit AI leverage. |
| Market Timing | 7 | Post-COVID telehealth acceptance is high. Nursing home staffing crises are severe. CMS has expanded telehealth reimbursement. But the regulatory and liability landscape remains challenging. |
| Small Team Fit | 3 | Requires licensed physicians, liability insurance, state-by-state licensing, 24/7 coverage, and healthcare compliance infrastructure. This is fundamentally not a 5-person operation. |
| Revenue Clarity | 6 | Reimbursement pathways exist but are complex. Per-facility subscription possible. But unit economics require significant clinical staff costs. |
| Defensibility | 5 | Clinical relationships and outcome data provide some moat. But telehealth is commoditized and larger players dominate. |
Weighted Total: 5.35
TIER: AI-Powered Sales & GTM¶
6. CoffeeAI (W22) -- AI Personalized Sales Outreach¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | LLMs are exceptional at personalization at scale. Researching prospects, crafting contextual messages, and iterating on tone/angle is precisely what modern AI does well. The product Opkit dreamed of in 2022 is trivially buildable now. |
| Market Timing | 6 | The market is saturated. Clay, Instantly, Apollo, Smartlead, Lavender, and dozens of others occupy this space. Email deliverability is harder than ever due to Google/Yahoo anti-spam changes. Cold outreach fatigue is real. |
| Small Team Fit | 9 | Pure software, no compliance overhead, fast iteration cycles. A small team can build and ship quickly. |
| Revenue Clarity | 8 | SaaS pricing, self-serve possible, clear value prop. But pricing pressure is intense due to competition. |
| Defensibility | 3 | Almost zero moat. The feature set is easily replicated. Every CRM and sales engagement platform is adding AI personalization. This is a feature, not a company. |
Weighted Total: 7.05
7. Flike (W22) -- AI Sales Email Copywriting¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | LLMs write excellent sales copy. Fine-tuning on winning emails improves performance. But this is now a commodity capability -- ChatGPT does this adequately. |
| Market Timing | 4 | This is a solved problem. Every sales tool has AI email writing built in. Standalone AI copywriting tools have been subsumed by platforms. The window has closed. |
| Small Team Fit | 9 | Simple product, easy to build. |
| Revenue Clarity | 5 | Hard to charge meaningful amounts for what is now a feature in every sales platform. |
| Defensibility | 2 | No moat whatsoever. This is a feature in Salesforce, HubSpot, Outreach, Salesloft, and every other sales tool. |
Weighted Total: 5.60
8. Fabius (W23) -- AI Sales Call Optimization¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | Modern AI can transcribe calls perfectly, analyze sentiment, detect buying signals, score rep performance, and generate coaching recommendations. Real-time call assistance is now possible with low-latency models. |
| Market Timing | 7 | Revenue intelligence is a proven category (Gong at $250M+ ARR, Chorus acquired by ZoomInfo). But incumbents are powerful. Opportunity exists in the SMB/mid-market gap and in AI-native real-time coaching. |
| Small Team Fit | 7 | Core product is buildable. Integrations with dialers, CRMs, and video platforms add scope. But modern APIs make this manageable. |
| Revenue Clarity | 8 | Per-seat SaaS pricing. Clear ROI: improved win rates, shorter sales cycles, faster rep ramp. Proven willingness to pay in this category. |
| Defensibility | 6 | Proprietary scoring models trained on call outcomes create some moat. But Gong/Chorus have massive data advantages. Best defensibility comes from going deeper in a specific vertical or deal type. |
Weighted Total: 7.50
9. Buzzle (S21) -- Sales Conversation Analytics¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | Same AI capabilities as Fabius -- transcription, NLP, sentiment analysis are dramatically better now. |
| Market Timing | 6 | Gong dominates. The pure analytics play is harder to differentiate. Market is mature. |
| Small Team Fit | 7 | Buildable but the analytics layer requires significant data infrastructure. |
| Revenue Clarity | 7 | Proven category, proven pricing. But competing with Gong on analytics alone is tough. |
| Defensibility | 4 | Analytics on conversations is a feature Gong already owns. Hard to build independent moat. |
Weighted Total: 6.50
10. Demo Gorilla (W22) -- SaaS Demo Optimization¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI can now generate personalized demo environments, create interactive product tours automatically, analyze demo engagement, and provide real-time coaching. Synthetic demo data generation is trivial with LLMs. |
| Market Timing | 8 | Product-led growth is mainstream. Interactive demos are replacing live demos for initial qualification. Navattic, Storylane, and Walnut have validated the market but none are AI-native. Buyers want self-serve demo experiences. |
| Small Team Fit | 8 | Browser-based product, clear scope. Can start with one platform (web apps) and expand. |
| Revenue Clarity | 8 | Per-seat or per-demo SaaS. Clear buyer (sales/marketing leaders). Measurable ROI (demo-to-meeting conversion). |
| Defensibility | 6 | AI-generated demo environments with real product data could be differentiated. Integration depth with specific SaaS platforms creates switching costs. But incumbents can add AI features. |
Weighted Total: 7.60
11. CustomerOS (S22) -- B2B Lead Intelligence¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI improves data enrichment, intent signal detection, and lead scoring. But data acquisition and accuracy remain the hard problem, and AI alone does not solve it. |
| Market Timing | 6 | ZoomInfo, Apollo, Clearbit (now HubSpot), and Clay dominate. The market is crowded and consolidating. |
| Small Team Fit | 6 | Data products require significant data infrastructure, web scraping, and data quality management. Ongoing data freshness is operationally intensive. |
| Revenue Clarity | 7 | Proven pricing models. But competing on data quality against established players is expensive. |
| Defensibility | 4 | Data moats require scale that 5 people cannot achieve against ZoomInfo's 3000+ employees. |
Weighted Total: 6.10
12. Abbot (S21) -- Customer Success AI Copilot¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI can now synthesize customer health signals across email, support tickets, product usage, and CRM data. LLMs generate renewal playbooks, churn risk explanations, and expansion opportunity briefs. Agentic workflows can automate follow-ups. |
| Market Timing | 8 | Customer success is under pressure to prove ROI. CS teams are being cut. AI copilots that make 1 CSM as effective as 3 are in high demand. Gainsight and Totango are legacy platforms ripe for AI-native disruption. |
| Small Team Fit | 7 | Requires integrations with many data sources (CRM, support, product analytics, billing). But modern integration platforms (Merge, Nango) help. Core AI logic is manageable. |
| Revenue Clarity | 8 | Per-CSM seat pricing. Clear ROI: reduced churn, more expansion revenue, fewer CSMs needed. Budget exists (CS teams already spend on Gainsight-class tools). |
| Defensibility | 6 | Customer health scoring models improve with data. Workflow customization creates switching costs. But Gainsight will add AI features aggressively. |
Weighted Total: 7.50
TIER: AI Developer Tools¶
13. CodeStory/Aide (S23) -- AI-Native IDE¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 10 | This IS an AI product. The entire value prop is AI-powered coding. Modern LLMs (Claude, GPT-4, Gemini) make AI-assisted development genuinely transformative. Agentic coding is the hottest space in AI. |
| Market Timing | 5 | The market is white-hot but dominated by Cursor, Windsurf, GitHub Copilot, and others with massive funding and user bases. Claude Code and similar agent-mode tools are raising the bar weekly. Extremely difficult to differentiate. |
| Small Team Fit | 6 | Building an IDE is a massive undertaking. VS Code extensions are easier but limit differentiation. Keeping up with the pace of model improvements and competitor features is exhausting for a small team. |
| Revenue Clarity | 7 | Subscription pricing proven by Cursor ($20/mo). Developers will pay. But customer acquisition against Cursor's brand is expensive. |
| Defensibility | 3 | Almost impossible to defend. Model improvements benefit all players equally. IDE features are quickly copied. The leaders have orders of magnitude more funding and users. |
Weighted Total: 6.35
14. Neptyne (W23) -- Programmable Spreadsheet with Code¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI can now generate complex spreadsheet formulas, write Python transformations from natural language, and automate data workflows. The "talk to your spreadsheet" paradigm is powerful. |
| Market Timing | 7 | Data teams are underserved. The gap between spreadsheets and code is a real pain point. However, Google Sheets has AI features, and tools like Rows.com and Equals are in market. |
| Small Team Fit | 6 | Building a spreadsheet engine is surprisingly complex. Real-time collaboration, formula evaluation, and data connectors are significant engineering challenges. |
| Revenue Clarity | 7 | Per-seat SaaS. Data teams will pay for better tooling. But spreadsheet tools have historically struggled to charge premium prices. |
| Defensibility | 5 | Template libraries and data connectors create some switching costs. But Google and Microsoft can replicate features. |
Weighted Total: 6.65
15. CodeParrot AI (W23) -- UI Dev Acceleration¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | AI-powered design-to-code (Figma to React/Vue/etc.) is dramatically better now. Vision models understand UI layouts. LLMs generate clean component code. This is a genuine 10x improvement over manual UI development. |
| Market Timing | 7 | Design-to-code is validated by v0, Bolt, and Lovable. But those tools are well-funded and gaining traction. The opportunity is in deeper integration with existing design workflows rather than greenfield generation. |
| Small Team Fit | 8 | Focused scope (UI development only). Can build as a Figma plugin or CLI tool. Less surface area than a full IDE. |
| Revenue Clarity | 7 | Per-seat or usage-based pricing. Frontend teams are large and underserved. But willingness to pay for dev tools varies. |
| Defensibility | 4 | Component libraries and design system understanding could create some moat. But the space is moving fast and well-funded competitors dominate. |
Weighted Total: 7.15
16. Sublingual (W25) -- Developer Productivity Tracking¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI can analyze code commits, PR patterns, and development workflows to surface productivity insights without invasive monitoring. Sentiment analysis on communications adds context. |
| Market Timing | 6 | Developer productivity measurement is politically sensitive. DORA metrics are standard but limited. Engineering leaders want better data but developers resist surveillance. Post-layoff era makes this more relevant but also more contentious. |
| Small Team Fit | 8 | Data pipeline from Git/GitHub/Jira plus AI analysis layer. Manageable scope. |
| Revenue Clarity | 6 | Engineering leaders will pay, but the buyer is often skeptical of the category. Pricing is per-developer, which creates pushback. |
| Defensibility | 5 | Benchmarking data across companies could be valuable. But LinearB, Jellyfish, and others are established. |
Weighted Total: 6.40
TIER: Legal / Compliance / Documents¶
17. Abel (W24) -- AI Document Review for Litigation¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | LLMs are exceptional at document review -- reading, classifying, extracting key provisions, identifying relevance, and flagging privilege. What required armies of contract attorneys can now be done by AI with human oversight. |
| Market Timing | 9 | Legal AI adoption has crossed the chasm. AmLaw 100 firms are actively deploying AI. E-discovery and document review is a $12B+ market with proven willingness to pay. Courts are increasingly accepting AI-assisted review. |
| Small Team Fit | 6 | Legal compliance, data security requirements (client-attorney privilege), and the need for law firm sales relationships add overhead. On-premise deployment requirements for some firms are burdensome. |
| Revenue Clarity | 9 | Per-document or per-matter pricing. Legal budgets are enormous. Document review currently costs $25-75/hour for contract attorneys -- AI can do it for pennies. ROI is immediate and massive. |
| Defensibility | 6 | Legal-specific fine-tuning, case law knowledge, and court-specific formatting create some moat. But Relativity, Everlaw, and other e-discovery platforms are adding AI aggressively. |
Weighted Total: 7.85
18. Dialect (S22) -- Generative AI Form/RFX Assistant¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | RFP/RFQ/RFI response is a perfect AI use case. LLMs can ingest past responses, company knowledge bases, and product documentation to generate tailored proposal content. This was barely possible in 2022; it is excellent now. |
| Market Timing | 8 | Enterprise procurement is increasingly RFP-driven. Companies waste enormous time on proposal responses. Loopio and Responsive (formerly RFPIO) have validated the market, but neither is AI-native. |
| Small Team Fit | 8 | Clear product scope: ingest company knowledge, match to RFP questions, generate responses, enable human review. No regulatory burden. Standard B2B SaaS. |
| Revenue Clarity | 9 | Per-user or per-proposal SaaS. Sales/presales teams are the buyer. Clear ROI: 80% reduction in proposal writing time. Large enterprises respond to hundreds of RFPs per year. |
| Defensibility | 7 | The company-specific knowledge base becomes the moat. Past responses, win/loss data, and institutional knowledge compound over time. Switching costs are high once the knowledge base is built. |
Weighted Total: 8.25
19. Telivy (S21) -- Cybersecurity Risk Assessment¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI can analyze security configurations, scan for vulnerabilities, and generate risk reports. But cybersecurity assessment still requires technical depth that AI augments rather than replaces. |
| Market Timing | 8 | Cyber insurance requirements are driving demand for risk assessments. SEC cyber disclosure rules create compliance pressure. SMBs need affordable security assessments but cannot afford traditional consultants. |
| Small Team Fit | 6 | Security tools require deep technical expertise, ongoing vulnerability database maintenance, and scanner infrastructure. Compliance certifications (SOC2, etc.) are important for credibility. |
| Revenue Clarity | 8 | Per-assessment or subscription pricing. Cyber insurance brokers are a distribution channel. SMB market is price-sensitive but the volume is enormous. |
| Defensibility | 6 | Risk scoring models and industry benchmarking data create some moat. But the cybersecurity market is extremely competitive. |
Weighted Total: 7.05
TIER: Data & Analytics¶
20. Vizly (S23) -- Data to Insights Platform¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | "Talk to your data" is a killer AI use case. LLMs can write SQL, generate visualizations, explain trends, and surface anomalies from natural language questions. Code interpreters make this robust. |
| Market Timing | 7 | The space is competitive: ThoughtSpot, Mode, Hex, and AI features in Tableau/Looker. But none are truly AI-native. The gap between "business person has a question" and "analyst writes SQL" remains huge. |
| Small Team Fit | 7 | Core product (natural language to SQL/charts) is buildable. But data connectors, security, and handling diverse data schemas add complexity. |
| Revenue Clarity | 7 | Per-seat SaaS for data teams and business users. But BI tool pricing is under pressure. Must be meaningfully cheaper or better than existing tools. |
| Defensibility | 5 | Schema understanding and query patterns could create some advantage. But this is a feature every BI tool is adding. |
Weighted Total: 7.15
21. Mercator (S22) -- AI-Assisted Data Analytics¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | Similar to Vizly -- AI dramatically improves data analytics accessibility. |
| Market Timing | 6 | Same competitive landscape as Vizly but with less differentiation. |
| Small Team Fit | 7 | Manageable scope if focused on a specific data domain. |
| Revenue Clarity | 6 | Harder to differentiate pricing against established BI tools. |
| Defensibility | 4 | Generic data analytics has no moat against incumbents adding AI. |
Weighted Total: 6.30
22. Orbiter (W20) -- ML Business Performance Monitoring¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | Anomaly detection and root cause analysis benefit from AI. But ML monitoring was already the core product -- the AI improvement is incremental, not transformative. |
| Market Timing | 6 | Business monitoring is served by Datadog, Monte Carlo, and others. The market exists but is competitive. |
| Small Team Fit | 6 | Data pipeline infrastructure and real-time monitoring are operationally intensive. |
| Revenue Clarity | 7 | Usage-based or per-data-source pricing. Proven category. |
| Defensibility | 4 | Hard to defend against Datadog and other monitoring platforms adding AI features. |
Weighted Total: 6.10
23. Tydo (S20) -- Data Intelligence¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI improves data aggregation and insight generation. But "data intelligence" is vague -- the AI leverage depends on specific use case focus. |
| Market Timing | 5 | Generic data intelligence is not a compelling category. Specific verticals (e-commerce analytics, marketing analytics) are better positioned. |
| Small Team Fit | 7 | Depends on scope. Narrow focus is buildable. |
| Revenue Clarity | 5 | Unclear without specific vertical focus. |
| Defensibility | 3 | No moat in generic data intelligence. |
Weighted Total: 5.55
TIER: Healthcare AI¶
24. RadMate AI (W24) -- Radiologist AI Copilot¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | Vision AI for radiology has made enormous strides. AI can now detect findings, generate structured reports, prioritize worklists, and provide differential diagnoses. Multi-modal models combining images with clinical context are powerful. |
| Market Timing | 8 | Radiologist shortage is severe and worsening. Burnout is epidemic. ACR has endorsed AI adoption. FDA has cleared 700+ AI radiology algorithms. Hospitals are actively budgeting for AI radiology tools. |
| Small Team Fit | 4 | FDA clearance (510(k)) is required for clinical diagnostic use. This takes 12-18 months minimum and significant capital. HIPAA compliance, DICOM integration, and clinical validation studies are mandatory. |
| Revenue Clarity | 8 | Per-study or per-radiologist pricing. Hospitals pay $100-400K/year for radiology AI tools. Clear budget line item. |
| Defensibility | 7 | Clinical validation data, FDA clearances, and integration with PACS/RIS systems create strong moats. But competition from Aidoc, Viz.ai, and others is intense. |
Weighted Total: 7.25
25. Reverie Labs (W18) -- Drug Discovery AI¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI-driven molecular design, binding prediction, and ADMET optimization are transformative. AlphaFold and successors have changed structural biology. Generative chemistry is producing novel candidates. |
| Market Timing | 7 | Pharma is investing heavily in AI drug discovery partnerships. But the space has been over-hyped and some early AI drug discovery companies have failed to deliver clinical results. Market is maturing. |
| Small Team Fit | 2 | Drug discovery requires wet lab validation, medicinal chemistry expertise, biology knowledge, and eventually clinical trials. This is fundamentally not a software-only or small-team endeavor. Capital requirements are enormous. |
| Revenue Clarity | 5 | Partnership/licensing model with pharma. Long timelines to revenue. Milestone payments are uncertain. |
| Defensibility | 6 | Proprietary models and molecular libraries create some IP. But the competitive landscape (Recursion, Insilico, Isomorphic Labs) is formidable. |
Weighted Total: 5.65
26. Clear Genetics (W17) -- Genomic Services¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI can interpret genomic variants, generate patient-friendly reports, and automate genetic counseling triage. But clinical genetics requires licensed counselors for many interactions. |
| Market Timing | 7 | Genetic testing is growing rapidly. Direct-to-consumer genomics has increased demand for interpretation. Shortage of genetic counselors creates opportunity for AI augmentation. |
| Small Team Fit | 4 | Requires licensed genetic counselors, CLIA/CAP compliance if offering clinical services, and integration with lab systems. Regulatory overhead is significant. |
| Revenue Clarity | 6 | Per-test interpretation or subscription for genetic counseling services. But reimbursement for genetic counseling is complex. |
| Defensibility | 5 | Variant interpretation databases improve with scale. But large lab companies (Invitae, before bankruptcy, and others) have more data. |
Weighted Total: 5.85
TIER: Vertical AI¶
27. Nophin (W22) -- Commercial Real Estate AI Deal Screening¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI can analyze offering memorandums, extract key financial metrics, compare to market comps, and score deals automatically. LLMs excel at parsing complex real estate documents. Vision models can analyze floor plans and property photos. |
| Market Timing | 6 | Commercial real estate is cyclical and currently in a challenging period (office vacancies, interest rates). But distressed asset opportunities and the need for faster deal screening persist. |
| Small Team Fit | 8 | Document analysis plus financial modeling is a manageable scope. No regulatory burden beyond standard data security. Clear, narrow use case. |
| Revenue Clarity | 8 | Per-deal or subscription pricing for CRE firms, private equity, and REITs. These buyers have large budgets and clear pain points. Deal volume justifies the investment. |
| Defensibility | 6 | CRE-specific document understanding, market comp databases, and deal outcome data create compounding advantage. Relatively niche market limits big tech competition. |
Weighted Total: 7.25
28. Onsite Pro (S21) -- Digital Sales for HVAC¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI can generate proposals from photos of existing systems, calculate load requirements, recommend equipment, and optimize pricing. Computer vision for site assessment is improving. |
| Market Timing | 7 | HVAC market is large ($25B+ residential) and fragmented. Digital transformation in trades is behind other industries. Electrification trends (heat pumps) are creating new sales complexity. |
| Small Team Fit | 7 | Narrow vertical focus is good. But HVAC-specific knowledge (Manual J calculations, equipment databases, local codes) requires domain expertise. |
| Revenue Clarity | 8 | Per-technician or per-company SaaS. HVAC companies are used to paying for software (ServiceTitan, Housecall Pro). Clear ROI: faster proposals, higher close rates. |
| Defensibility | 7 | Deep HVAC-specific models, equipment databases, and pricing intelligence create vertical moat. Incumbents (ServiceTitan) focus on operations, not AI-powered sales. |
Weighted Total: 7.15
29. Jasmine Energy (S22) -- Solar Incentive Automation¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI can track changing incentive programs, match properties to eligible incentives, and automate application paperwork. LLMs handle form filling and document preparation well. |
| Market Timing | 8 | IRA (Inflation Reduction Act) created massive new solar incentives. The complexity of federal, state, and local incentive stacking is overwhelming for installers. Demand for solar continues to grow despite policy uncertainty. |
| Small Team Fit | 7 | Regulatory tracking and form automation are manageable. But staying current with constantly changing incentive programs is operationally intensive. |
| Revenue Clarity | 8 | Per-project or subscription pricing for solar installers. Can also take a percentage of incentives captured. Clear ROI: installers leave money on the table without proper incentive optimization. |
| Defensibility | 7 | Comprehensive incentive database with real-time updates is hard to replicate. Success rate data and application optimization create compounding advantage. Policy complexity is a natural barrier. |
Weighted Total: 7.35
30. AI.Fashion (S20) -- AI Creative Suite for Fashion¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | Generative AI for fashion imagery is revolutionary. Virtual try-on, model generation, background replacement, and style transfer are all dramatically better. Diffusion models handle fashion aesthetics well. |
| Market Timing | 8 | E-commerce brands need massive amounts of product imagery. Fashion photography is expensive ($500-5000 per look). AI-generated models and virtual try-on are gaining acceptance. Sustainability pressure reduces desire for wasteful photoshoots. |
| Small Team Fit | 8 | Image generation pipeline plus web interface. Can leverage existing models (Stable Diffusion, DALL-E, Flux) and fine-tune for fashion. Clear, focused product. |
| Revenue Clarity | 8 | Per-image or subscription pricing. Fashion brands and e-commerce companies are the clear buyer. Replaces expensive photography budgets with 90%+ cost savings. |
| Defensibility | 6 | Fashion-specific fine-tuned models, brand style memory, and garment understanding create some moat. But generative AI image quality is improving for everyone. |
Weighted Total: 7.85
TIER: Productivity & Knowledge¶
31. Cardinal (W23) -- AI Product Backlog Management¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI can auto-generate user stories from conversations, deduplicate backlog items, score priority based on customer feedback signals, and link requests to strategic objectives. LLMs are excellent at synthesizing qualitative feedback into structured requirements. |
| Market Timing | 7 | Product teams are overwhelmed with feedback from multiple channels. The gap between customer insight and backlog management is real. But Jira, Linear, and Productboard are entrenched. |
| Small Team Fit | 8 | Clear product scope. Integration with existing tools (Jira, Linear, Slack, Intercom) via APIs. No regulatory overhead. |
| Revenue Clarity | 7 | Per-PM seat pricing. Product teams are accustomed to paying for tools. But must justify over Jira/Linear + AI features being added to those platforms. |
| Defensibility | 5 | Customer feedback pattern data could compound. But this is a feature that Jira/Linear/Productboard will add. |
Weighted Total: 7.10
32. Station (W18) -- Team Knowledge Management¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI-powered search across all company tools, automatic knowledge extraction from conversations, and proactive knowledge surfacing are transformative for knowledge management. RAG makes this viable. |
| Market Timing | 7 | Knowledge management pain is universal. Notion, Confluence, and Slite are entrenched but none are truly AI-native. Glean has validated the AI-powered enterprise search category at scale. |
| Small Team Fit | 6 | Integration with many tools (Slack, Google Drive, Notion, Confluence, etc.) is required. Security and permissions management across sources is complex. |
| Revenue Clarity | 7 | Per-seat SaaS. Teams pay for knowledge management already. But Glean is well-funded and dominant in enterprise. |
| Defensibility | 5 | Company-specific knowledge graphs create switching costs. But Glean has a massive head start with $350M+ raised. |
Weighted Total: 6.65
33. Mindmesh (S21) -- Virtual Workspace¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 6 | AI can enhance virtual workspaces with meeting summaries, task extraction, and smart scheduling. But the core value of "virtual workspace" is collaboration infrastructure, not AI. |
| Market Timing | 5 | Remote work tools are mature. Slack, Teams, Notion, and Linear dominate. The "virtual workspace" concept has not found product-market fit despite many attempts. |
| Small Team Fit | 5 | Building collaboration infrastructure is enormously complex. Real-time sync, permissions, integrations, and reliability are all table stakes. |
| Revenue Clarity | 5 | Per-seat SaaS, but the market is saturated and users have tool fatigue. |
| Defensibility | 3 | No moat against Slack, Teams, and Notion. |
Weighted Total: 4.95
34. Sorted (W23) -- SaaS Management¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 7 | AI can discover shadow IT, analyze usage patterns, recommend consolidation, and negotiate renewals. But the core challenge is data integration, not AI. |
| Market Timing | 7 | SaaS sprawl is a real problem. CFOs are scrutinizing software spend. Zylo and Productiv have validated the category. |
| Small Team Fit | 7 | Integration with SSO providers, financial systems, and SaaS platforms. Manageable scope. |
| Revenue Clarity | 7 | Percentage of savings or subscription pricing. Clear ROI: companies typically save 20-30% on SaaS spend. |
| Defensibility | 5 | Pricing intelligence databases and benchmark data create some advantage. But the category has established players. |
Weighted Total: 6.60
TIER: Customer Support AI¶
35. Parabolic (W23) -- Customer Support AI Assistant¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 9 | AI can now resolve 60-80% of customer support tickets autonomously. LLMs understand context, access knowledge bases, and generate accurate, empathetic responses. This is one of the highest-leverage AI applications. |
| Market Timing | 7 | The market is validated but crowded. Intercom Fin, Zendesk AI, Ada, and Forethought are all in-market. However, most solutions are mediocre and customers are willing to switch for better resolution rates. |
| Small Team Fit | 8 | Core product is a knowledge-base-powered AI agent. Integrations with support platforms (Zendesk, Intercom, Freshdesk) are well-documented. Clear scope. |
| Revenue Clarity | 8 | Per-resolution or per-ticket pricing aligns incentives. Clear ROI: reduce support headcount by 50%+. Support budgets are large and growing. |
| Defensibility | 5 | Company-specific training data improves over time. But the space is extremely competitive and major platforms are building this in. |
Weighted Total: 7.50
36. Brevy (S20) -- Customer Service AI Automation¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | Same AI capabilities as Parabolic but from 2020 -- the gap between what was possible then and now is enormous. |
| Market Timing | 6 | Same competitive landscape as Parabolic but with less recent momentum. |
| Small Team Fit | 8 | Manageable scope. |
| Revenue Clarity | 7 | Proven pricing models. |
| Defensibility | 4 | Even less differentiation than Parabolic in a crowded market. |
Weighted Total: 6.65
TIER: Content & Creative¶
37. Booth AI (W23) -- AI Product Photography¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 10 | AI product photography is a solved problem in 2026. Image generation models produce photorealistic product shots, lifestyle imagery, and model compositions. Virtual try-on is mainstream. This is a 100x cost reduction over traditional photography. |
| Market Timing | 8 | E-commerce is growing. Product photography budgets are under pressure. Early adopters have proven AI-generated product images perform as well as traditional photography in conversion tests. Amazon and Shopify have built basic versions. |
| Small Team Fit | 8 | Image generation pipeline, web/API interface, template system. Can leverage existing models and fine-tune. Clear, focused product. |
| Revenue Clarity | 8 | Per-image or subscription pricing. E-commerce brands are the buyer. Clear ROI: $50-500 per traditional product photo vs. $0.50-5 for AI-generated. |
| Defensibility | 5 | Model fine-tuning on product categories and brand styles creates some advantage. But generative AI is commoditizing rapidly. Amazon and Shopify offering basic versions for free is a threat. |
Weighted Total: 7.95
38. CreatorML (W23) -- Human Attention Prediction¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI models can now predict engagement, virality, and attention patterns based on content analysis. Thumbnail optimization, title analysis, and content scoring are valuable for creators. |
| Market Timing | 7 | Creator economy is enormous. YouTube, TikTok, and Instagram creators desperately want to understand what drives engagement. But platform algorithms change frequently, making predictions unstable. |
| Small Team Fit | 7 | ML pipeline for content analysis plus web interface. Training data acquisition is the main challenge. |
| Revenue Clarity | 6 | Per-creator subscription. But individual creators are price-sensitive. Agencies and MCNs are better buyers. |
| Defensibility | 6 | Prediction model accuracy improves with data. Platform-specific insights are valuable. But platform API changes can break the product overnight. |
Weighted Total: 6.85
39. Ozone (W22) -- Figma for Video¶
| Dimension | Score | Rationale |
|---|---|---|
| AI Leverage | 8 | AI video editing is transformative: auto-cut, scene detection, voice-driven editing, AI B-roll generation, auto-captioning, and style transfer. Natural language video editing commands are now possible. |
| Market Timing | 7 | Video content demand is exploding. But the market has strong players: Descript, Runway, CapCut, and Adobe are all adding AI. Collaboration in video editing is still underserved though. |
| Small Team Fit | 5 | Video editing software is extremely complex: rendering pipelines, format support, real-time preview, collaboration infrastructure. This is a massive engineering challenge for 5 people. |
| Revenue Clarity | 7 | Per-seat subscription. Creative teams pay for video tools. But competing with free/cheap tools (CapCut) and professional tools (Adobe) is a squeeze. |
| Defensibility | 4 | Collaboration features could differentiate, but Descript and Adobe have more resources. AI video features are commoditizing rapidly. |
Weighted Total: 6.35
FINAL RANKED LIST: Top 20 Opportunities¶
| Rank | Startup | Weighted Score | Top Signal | Rebuild Rationale |
|---|---|---|---|---|
| 1 | Opkit (S21) | 8.75 | AI Leverage: 10 | Healthcare phone call automation is a perfect storm: voice AI is now exceptional, healthcare admin costs are exploding, staff shortages are acute, and the ROI is undeniable. The product that was impossible in 2021 is now buildable in weeks. Per-call pricing with 90%+ cost savings for practices. HIPAA is a moat, not just a burden -- it keeps casual competitors out. |
| 2 | Argovox (S22) | 8.50 | Revenue Clarity: 9 | Patient billing is the highest-value vertical within healthcare voice AI. Revenue cycle management companies are desperate to automate collections calls. The willingness to pay is extraordinary because the ROI is measured in actual dollars collected. Narrower scope than Opkit means faster time to market. Can price as percentage of collections improvement. |
| 3 | Dialect (S22) | 8.25 | Defensibility: 7 | RFP response automation is an AI sweet spot: structured input, knowledge-intensive, high-volume, and currently done by expensive humans. The company knowledge base becomes a compounding moat. Incumbents (Loopio, Responsive) are legacy software with AI bolted on. An AI-native approach wins on speed, quality, and cost. $50K+ ACV is realistic from day one. |
| 4 | Booth AI (W23) | 7.95 | AI Leverage: 10 | AI product photography is one of the clearest AI value propositions in existence. The cost reduction is 100x. The quality gap has closed. E-commerce brands need thousands of images and currently spend $50-500 per shot. The threat from Amazon/Shopify is real but their offerings are basic -- a specialized tool with brand consistency, batch processing, and e-commerce workflow integration wins. |
| 5 | Abel (W24) | 7.85 | Revenue Clarity: 9 | Legal document review is a $12B+ market where AI delivers 10-50x cost reduction. Law firms bill $25-75/hour for contract attorney review; AI does it for pennies. The regulatory environment has shifted to accept AI-assisted review. The main risk is sales cycle length for law firms, but the ROI story is irresistible. FDA-style clearance is not needed -- this is a productivity tool, not a legal practice tool. |
| 6 | AI.Fashion (S20) | 7.85 | AI Leverage: 9 | Fashion-specific generative AI is differentiated from generic product photography. Virtual try-on, model diversity, style-consistent collections, and lookbook generation are high-value capabilities. Fashion brands pay premium prices for creative tools. The vertical focus creates defensibility against generic image generation tools. |
| 7 | Demo Gorilla (W22) | 7.60 | Market Timing: 8 | Interactive, personalized product demos are replacing live demos for initial qualification. Navattic and Storylane have validated the market but none are AI-native. AI can generate demo environments automatically, personalize for each prospect, and analyze engagement. The buyer (sales/marketing leaders) has clear budget and clear pain. |
| 8 | Fabius (W23) | 7.50 | AI Leverage: 9 | AI-native sales call intelligence is the right play against Gong. Real-time coaching (not just post-call analytics), automatic CRM updates, and deal risk scoring are differentiated. The SMB/mid-market segment cannot afford Gong's pricing. An AI-native tool at 1/3 the price with better real-time features wins. |
| 9 | Abbot (S21) | 7.50 | Market Timing: 8 | Customer success teams are being asked to do more with less. AI copilots that synthesize health signals across all touchpoints and generate proactive playbooks are in high demand. Gainsight is legacy software. An AI-native CS platform that makes 1 CSM as effective as 3 has immediate product-market fit. |
| 10 | Parabolic (W23) | 7.50 | AI Leverage: 9 | AI customer support resolution rates have reached 60-80% for well-implemented systems. The market is competitive but most solutions are mediocre. A team obsessed with resolution quality, not just deflection metrics, can win. Per-resolution pricing aligns incentives perfectly. The key differentiation is accuracy, not features. |
| 11 | Jasmine Energy (S22) | 7.35 | Defensibility: 7 | Solar incentive complexity is a natural moat. The IRA created a maze of federal, state, and local incentives that solar installers cannot navigate alone. An AI system that tracks every incentive program, matches properties to eligibility, and automates applications is worth thousands per project. Policy complexity ensures this cannot be a feature in existing solar software. |
| 12 | Struct (W23) | 7.35 | AI Leverage: 9 | Multi-lingual voice AI is a massive market but requires vertical focus to defend. The play is to own a specific vertical (e.g., multi-lingual healthcare, multi-lingual customer support for specific industries) rather than being a horizontal voice platform. The technology is ready; the strategy must be narrow. |
| 13 | Nophin (W22) | 7.25 | Small Team Fit: 8 | CRE deal screening is a focused, high-value use case. Private equity firms and REITs review hundreds of offering memorandums per quarter. AI can extract, analyze, and score deals in minutes vs. hours. The buyer has budget and pain. CRE-specific document understanding is genuinely specialized. Market cyclicality is a risk but distressed opportunities create demand in any cycle. |
| 14 | RadMate AI (W24) | 7.25 | Market Timing: 8 | Radiologist AI is a massive market with severe talent shortages. The technology works. But FDA clearance requirements make this very hard for a 5-person team. Best approached as a workflow/productivity tool (report generation, worklist prioritization) rather than a diagnostic tool, avoiding FDA requirements while still delivering enormous value. |
| 15 | CodeParrot AI (W23) | 7.15 | AI Leverage: 9 | Design-to-code is a focused niche within the broader AI dev tools space. Rather than competing with Cursor on general coding, owning the Figma-to-production-component pipeline is defensible. Frontend teams are large, the pain is real, and the existing solutions (v0, Bolt) focus on greenfield generation rather than design system compliance. |
| 16 | Vizly (S23) | 7.15 | AI Leverage: 9 | "Talk to your data" remains one of the most compelling AI use cases. The key is vertical focus: rather than competing with Tableau on general BI, own a specific data domain (e-commerce analytics, marketing performance, financial reporting). Vertical-specific data models and benchmarks create defensibility. |
| 17 | Onsite Pro (S21) | 7.15 | Defensibility: 7 | HVAC sales optimization is a genuinely underserved vertical. AI-powered site assessment from photos, automatic load calculations, equipment recommendations, and instant proposal generation save hours per job. The vertical expertise required keeps generic AI tools out. Heat pump transition creates new complexity and demand. |
| 18 | Cardinal (W23) | 7.10 | Small Team Fit: 8 | AI product backlog management solves a real pain point for PMs drowning in feedback from Slack, Intercom, sales calls, and support tickets. The risk is becoming a feature in Jira/Linear. The play is to own the "customer voice to product decision" workflow end-to-end, not just backlog management. Speed to market matters -- this feature gap will close. |
| 19 | Telivy (S21) | 7.05 | Market Timing: 8 | Cybersecurity risk assessment demand is driven by cyber insurance requirements and SEC disclosure rules. SMBs need affordable assessments. AI can automate 80% of a security audit that currently costs $10-50K from consultants. The insurance distribution channel is powerful. |
| 20 | CoffeeAI (W22) | 7.05 | AI Leverage: 9 | AI-powered sales outreach personalization is technically excellent and easy to build. The challenge is extreme market saturation. The only winning strategy is deep vertical specialization (e.g., AI outreach for healthcare, AI outreach for real estate) or a fundamentally different approach (fully autonomous SDR agent, not just email personalization). |
Dark Horse Opportunities¶
These startups score lower overall but have exceptional potential in one dimension that could make them breakout opportunities under the right conditions.
1. RadMate AI (W24) -- Radiologist AI Copilot¶
Dark Horse Factor: AI Leverage (9) + Market Timing (8)
The radiologist shortage is a genuine crisis. Burnout rates exceed 50%. AI that handles routine reads and generates structured reports can make each radiologist 2-3x more productive. The technology is proven. The reason this ranks as a dark horse rather than top-5 is the FDA barrier -- clinical diagnostic claims require 510(k) clearance. However, a clever team can position as a workflow and documentation tool (report drafting, prior comparison, worklist triaging) rather than a diagnostic tool, sidestepping FDA while still delivering massive value. If a team has a radiologist co-founder who understands this regulatory arbitrage, this becomes a top-5 opportunity. The market is $2B+ and the buyers are desperate.
2. Reverie Labs (W18) -- Drug Discovery AI¶
Dark Horse Factor: Revenue Clarity via Platform Pivot
Drug discovery itself is a terrible small-team opportunity (score: 5.65). But the underlying capability -- AI-driven molecular optimization and compound screening -- can be repackaged as a platform for contract research organizations (CROs) and academic labs. Rather than discovering drugs, sell the picks and shovels. A SaaS tool that helps medicinal chemists explore chemical space, predict ADMET properties, and optimize lead compounds has clear per-seat pricing, no clinical trial risk, and a large addressable market. This pivot transforms a 2/10 small-team-fit score into a 7/10.
3. CreatorML (W23) -- Human Attention Prediction¶
Dark Horse Factor: Defensibility (6) + Unique Data Moat Potential
Attention prediction sounds like a nice-to-have, but the underlying capability -- predicting what content will perform before publication -- is enormously valuable. The dark horse play is to pivot from creator tools to advertising and media buying. Predicting ad creative performance before spending media budget is worth billions to the advertising industry. A model trained on content performance data across platforms becomes more accurate with every prediction, creating a genuine data flywheel. If positioned as an AI creative testing platform for ad agencies and brands, the revenue clarity jumps from 6 to 9. The $600B global ad market pays handsomely for even marginal improvements in creative performance prediction.
4. Ozone (W22) -- Figma for Video¶
Dark Horse Factor: Market Timing (7) + Massive Latent Demand
Video editing for teams is genuinely unsolved. Every company now produces video content (social media, product demos, training, marketing) but professional video editing tools are built for solo editors, not collaborative teams. The dark horse thesis is that AI reduces the skill floor for video editing so dramatically that non-editors can produce professional video. If the team focuses not on competing with Premiere/DaVinci but on enabling marketing teams and product teams to create video without editors -- using AI for cuts, transitions, audio cleanup, captions, and B-roll -- the market is enormous. The scoring suffers from engineering complexity (5 for small team fit), but strategic use of AI video APIs (Runway, Pika) and a thin collaboration layer on top could make this feasible.
5. Nophin (W22) -- Commercial Real Estate AI Deal Screening¶
Dark Horse Factor: Revenue Clarity (8) + Extreme Vertical Depth
CRE deal screening already ranks in the top 20, but it has dark horse potential because the opportunity is larger than it appears. The initial product (OM analysis and deal scoring) is a wedge into the entire CRE investment workflow: underwriting, due diligence, portfolio monitoring, and investor reporting. Each expansion doubles the ACV. CRE firms pay $50-200K/year for analytics platforms. With interest rates normalizing and transaction volumes recovering, the timing is improving. A team with one CRE industry insider can build relationships that no horizontal AI tool can replicate. The niche looks small but $20T in US commercial real estate assets generate enormous software budgets.
Strategic Recommendations¶
If building ONE company, build Opkit (Healthcare AI Phone Calls)¶
The combination of transformative AI leverage, urgent market need, clear revenue model, and regulatory moat makes this the highest-conviction opportunity. The original Opkit failed because the technology was not ready in 2021. In 2026, voice AI is exceptional, healthcare staffing shortages are worse, and the ROI case is undeniable.
Sprint Plan for Opkit Rebuild: - Week 1-2: Insurance verification call automation for a single payer (e.g., Aetna) - Week 3-4: Expand to 3-5 major payers, add prior authorization calls - Week 5-6: Launch with 5-10 pilot practices, iterate on call success rates - Months 2-3: Appointment scheduling, referral coordination - Months 4-6: Scale to 50+ practices, add billing company channel
If hedging, pair a top-5 with a dark horse¶
Run Dialect (RFP automation, rank 3) as the revenue engine -- it has the fastest path to meaningful revenue with the least technical risk -- while investing 20% of capacity in RadMate (radiology AI copilot) as the potential billion-dollar outcome.
Avoid the traps¶
- CodeStory/Aide: The AI IDE market is a knife fight with billion-dollar combatants. Do not enter.
- CoffeeAI/Flike: AI sales email is a commodity. The window has closed.
- Mindmesh/Station: Collaboration tools require massive scale to succeed. Not a small-team play.
- Reverie Labs (as drug discovery): Without wet labs and $50M+, do not attempt drug discovery.
- Call9: Emergency medicine requires clinical staff, liability coverage, and 24/7 operations. Not a software business.
Scoring Summary Table¶
| Rank | Startup | AI Lev. | Mkt Tim. | Sm Team | Rev Clar. | Defens. | Weighted |
|---|---|---|---|---|---|---|---|
| 1 | Opkit | 10 | 9 | 8 | 9 | 7 | 8.75 |
| 2 | Argovox | 9 | 9 | 8 | 9 | 7 | 8.50 |
| 3 | Dialect | 9 | 8 | 8 | 9 | 7 | 8.25 |
| 4 | Booth AI | 10 | 8 | 8 | 8 | 5 | 7.95 |
| 5 | Abel | 9 | 9 | 6 | 9 | 6 | 7.85 |
| 6 | AI.Fashion | 9 | 8 | 8 | 8 | 6 | 7.85 |
| 7 | Demo Gorilla | 8 | 8 | 8 | 8 | 6 | 7.60 |
| 8 | Fabius | 9 | 7 | 7 | 8 | 6 | 7.50 |
| 9 | Abbot | 8 | 8 | 7 | 8 | 6 | 7.50 |
| 10 | Parabolic | 9 | 7 | 8 | 8 | 5 | 7.50 |
| 11 | Jasmine Energy | 7 | 8 | 7 | 8 | 7 | 7.35 |
| 12 | Struct | 9 | 7 | 7 | 8 | 5 | 7.35 |
| 13 | Nophin | 8 | 6 | 8 | 8 | 6 | 7.25 |
| 14 | RadMate AI | 9 | 8 | 4 | 8 | 7 | 7.25 |
| 15 | CodeParrot AI | 9 | 7 | 8 | 7 | 4 | 7.15 |
| 16 | Vizly | 9 | 7 | 7 | 7 | 5 | 7.15 |
| 17 | Onsite Pro | 7 | 7 | 7 | 8 | 7 | 7.15 |
| 18 | Cardinal | 8 | 7 | 8 | 7 | 5 | 7.10 |
| 19 | Telivy | 7 | 8 | 6 | 8 | 6 | 7.05 |
| 20 | CoffeeAI | 9 | 6 | 9 | 8 | 3 | 7.05 |
| 21 | CreatorML | 8 | 7 | 7 | 6 | 6 | 6.85 |
| 22 | Neptyne | 8 | 7 | 6 | 7 | 5 | 6.65 |
| 23 | Station | 8 | 7 | 6 | 7 | 5 | 6.65 |
| 24 | Brevy | 8 | 6 | 8 | 7 | 4 | 6.65 |
| 25 | Sorted | 7 | 7 | 7 | 7 | 5 | 6.60 |
| 26 | Buzzle | 8 | 6 | 7 | 7 | 4 | 6.50 |
| 27 | Sublingual | 7 | 6 | 8 | 6 | 5 | 6.40 |
| 28 | CodeStory/Aide | 10 | 5 | 6 | 7 | 3 | 6.35 |
| 29 | Ozone | 8 | 7 | 5 | 7 | 4 | 6.35 |
| 30 | Mercator | 8 | 6 | 7 | 6 | 4 | 6.30 |
| 31 | VOIQ | 8 | 6 | 5 | 7 | 4 | 6.20 |
| 32 | CustomerOS | 7 | 6 | 6 | 7 | 4 | 6.10 |
| 33 | Orbiter | 7 | 6 | 6 | 7 | 4 | 6.10 |
| 34 | Clear Genetics | 7 | 7 | 4 | 6 | 5 | 5.85 |
| 35 | Reverie Labs | 8 | 7 | 2 | 5 | 6 | 5.65 |
| 36 | Flike | 8 | 4 | 9 | 5 | 2 | 5.60 |
| 37 | Tydo | 7 | 5 | 7 | 5 | 3 | 5.55 |
| 38 | Call9 | 6 | 7 | 3 | 6 | 5 | 5.35 |
| 39 | Mindmesh | 6 | 5 | 5 | 5 | 3 | 4.95 |
Analysis prepared March 2026. Scoring reflects current AI capabilities, market conditions, and competitive landscape as of this date. All assessments assume a 5-person technical team with 6 months of runway and no domain-specific regulatory licenses at the outset.