Skip to content
Sign in
SWC 2026 FinalistGrand Final: Lunch Block

Promptend

Deterministic verifier that checks AI-generated TypeScript against its specification before it reaches production.

Moldova, EuropeAI & software, Verification of AI-generated codeMoldova – SWC national final (Startup Moldova Summit, Chișinău, 22 Apr. 2026) — win claimed by promptend.ai
Stage
UndeterminedNo documented AI
Product
PrototypeNone known
Funding raised
Undisclosedno funding
Founded
20251 years ago
Founders
1Evgheni
Global score
3.0Grade DRank 126 of 151

Overview

Verification tool for AI-generated code that checks the actual behavior of TypeScript code before it goes into production.

Traction

None known

Product in early access (appointment booking); no customers or users announced.

Market

AI trust, risk and security (AI TRiSM)

AI coding assistants are generating a growing share of production code, while review capacity has not grown; teams need guarantees that generated code does what the spec says. Buyers are engineering leaders and platform teams, often adopting bottom-up via developers and CI. LLM-based code review is probabilistic, leaving room for deterministic verification tools.

Addressable market

2025
$2.9B
2030
$7.4B
+21.6% a year

AI trust, risk and security (AI TRiSM): global market from $2.9bn in 2025 to $7.44bn in 2030 (CAGR 21.6%).

Customers

Engineering teams shipping TypeScript; platform/DevEx teams; AI coding-agent vendors

Who signs the cheque

VP Engineering / platform leads; developer tool budgets

Geographies

  • Moldova
  • Global (developer tools)

Drivers

  • Explosion of AI-generated code
  • Review bottlenecks and production incidents
  • Regulated industries demanding traceability

Headwinds

  • Crowded AI code-review market
  • Single-language scope (TypeScript)
  • Minimal team

Global AI TRiSM market from $2.9bn in 2025 (The Business Research Company) to $7.44bn in 2030 (Grand View, CAGR 21.6%).

Scalability 9/10. Software developer tool distributable worldwide on a self-service basis; constraint: competition from built-in coding assistants.

Market source: gii.tw

Technology

Category

Developer tool (program analysis)

No documented AI

Maturity

  1. R&D
  2. Prototype
  3. Pilot
  4. On the market
  5. Scaling

prototype

How it works

Deterministic analysis predicts code execution and checks it against the specification rather than relying on probabilistic LLM review. Developers review the specification instead of every line.

Moat

Potentially strong if the analysis engine is robust; program-analysis IP is hard to build, but no patents or benchmarks are cited.

Differentiation

Deterministic, spec-based verification vs LLM review bots.

Integration points

  • CI/CD pipelines
  • Pull-request workflows
  • AI coding agents

Depends on

  • TypeScript ecosystem
  • Spec authoring by developers

Use of AI

No use of AI mentioned

Product in detail

Deterministic analysis that predicts code execution and checks it against the specification (vs probabilistic LLM reviews); the developer reviews the spec rather than every line.

Business

Model
Not published; likely per-seat or per-repo developer SaaS.
Revenue logic
Per developer seat or per verified PR.
Go-to-market
Early access via appointment booking; developer-led adoption.

Next milestones

  1. Public benchmark vs LLM reviewers
  2. Integration in an AI coding agent's PR flow
  3. First paying design partners

Key risks

  • No traction
  • Minimal team
  • Competition

Opportunities

4 found in the network

Concrete moves for Promptend inside the network: who to sell to, sell through, build with or raise from, each with the reasoning, the evidence and the first step.

  • Co-development 2
  • Investors 1
  • Peers 1

Co-development2

An integration or joint product

Verification gate in an AI feedback-to-PR pipeline

Starko's AI builds accepted product changes all the way to a verified pull request; Promptend checks that generated TypeScript matches its spec. Embedding Promptend as Starko's verification step strengthens Starko's 'verified PR' promise and gives Promptend a production integration.

35%confidence

Evidence from the records

  • “Starko turns customer feedback from every channel into product recommendations, then has its AI build the accepted changes all the way to a verified pull request.”
  • “Verification tool for AI-generated code that checks the actual behavior of TypeScript code before it goes into production.”
If it works
A live usage stream for Promptend and fewer regressions for Starko's customers.
First step
Run Promptend on a sample of Starko-generated PRs and compare defect detection.

Spec-based verification for test automation

TatraSpace sells testing infrastructure (remote test bench, CI, AI agent) for embedded and flight software; deterministic spec verification complements hardware-in-the-loop testing and could be offered to its customers.

15%confidence

Evidence from the records

  • “firmware and flight software, automated testing/CI”
  • “Deterministic analysis that predicts code execution and checks it against the specification”
If it works
Joint offer for safety-critical software teams.
First step
Discuss whether TatraSpace's toolchain has TypeScript components to pilot on.

Investors1

A fund whose thesis fits

European pre-seed AI/devtools

Speedinvest backs companies at the earliest stages across Europe in AI & software; a deterministic verification engine for AI code is a category-defining pitch if benchmarks exist.

15%confidence

Evidence from the records

  • “Back companies at the earliest stages to build category-defining businesses.”
  • “Growth theme”
If it works
Pre-seed capital to build the team.
First step
Publish a benchmark repo, then apply via Speedinvest's website.

Peers1

Founders solving the same problem elsewhere

AI reliability peer exchange

Both address trust in AI outputs (Promptend for generated code, Neurath for model predictions) and sell to risk-averse engineering buyers; sharing go-to-market learnings on assurance products is useful.

15%confidence

Evidence from the records

  • “Real problem”
  • “Reliability infrastructure for critical AI (robotics, defense, medicine, industry)”
If it works
Shared buyer insights and possible joint content.
First step
Set a founder call on selling assurance tools.

Analyst reading of 21 Nov 2026, from the public record of the network. Method: analyst-ai.

Investors that fit

US investor with no profile in our data; joined Hardshell's $1.1M pre-seed in AI training-data security.

Moldovan AI-code verification — AI TRiSM.

25%confidence

Evidence from the records

  • “Hardshell”
  • “Verification of AI-generated code”
  • “Product in early access (appointment booking)”
Intro path
Founder-to-founder intro via Hardshell (network portfolio company of MoJo Ventures).
First step
Low priority.

Antler: a global early-stage VC that backs founders from inception, investing pre-seed to Series A in 30+ cities.

Moldovan verification of AI-generated code: a timely AI-infrastructure theme, but early-access with no users.

20%confidence

Evidence from the records

  • “Verification of AI-generated code”
  • “Moldova”
First step
Apply to an Antler residency to build traction.

Talent investor pairing aspiring founders before ideas; 626 companies created across London, Paris, Berlin, Singapore, Bangalore and Toronto.

Pre-traction AI-code verification in early access: the only stage where an EF-style restart could make sense. A formed team is a structural mismatch.

10%confidence

Evidence from the records

  • “Verification of AI-generated code”
  • “Prototype”
First step
Only if a founder seeks a cofounder through EF.

Investment analysis

3.0D

Grade D

Rank 126 of 151

Disruption4.0/10
Business model1.0/10
Market5.0/10
Risk(lower is better)9.0/10

Global score = mean of disruption, model, market and (11 − risk). Last updated 21 Nov 2026.

Not investable.

Disruption

4/10

Verification of AI-generated code: emerging category.

Business model

1/10

Early access, no users.

Market

5/10

TAM AI trust, risk and security (AI TRiSM) ~$2.91bn (base 4), strong scalability 9/10 (+1), no proven international expansion (0) → 5/10.

Risk

9/10

Lower is better

Very high risk: one-person team.

Strengths

  • Real problem
  • Growth theme

Risks

  • No traction
  • Minimal team

Funding

Undisclosed, no funding
  1. 2026

    prizenot disclosed (SWC final qualification)

    Startup World Cup — regional final (Moldova)

Amount used in the study. No funding round; prize not quantified.

Note. no public funding round found

Country (Moldova) and qualification filled in: the website states it won an SWC regional competition and is “pitching at the SF final”; a LinkedIn post (Olga Melniciuc) presents it as a Moldovan startup alongside PsychoTech.

Team

  • Evgheni Demcenco

    Main contact / likely founder (role not confirmed)

    Sole contact person on the website (email and appointment booking); Moldovan startup.

Sources

Regional final
Moldova – SWC national final (Startup Moldova Summit, Chișinău, 22 Apr. 2026) — win claimed by promptend.ai
Identification confidence
medium
Founding year: 2025 (low confidence)
First website archived in 2025, proxy (year the website existed, no confirmed founding date) web.archive.org
  1. 1promptend.ai/
  2. 2linkedin.com/posts/olgamelniciuc_psychotech-and-promptendai-two-moldovan
  3. 3umaef.org/medias/startup-moldova-summit-2026-showcases-moldovas-growi

From the open study Startup World Cup 2026 by TIBOK, based on public sources, verified in November 2026. Scores are an analyst's judgment on an explicit grid, not the jury's.

Wall

Wall

No posts yet