# Enkrypt AI

Aiolum story: https://aiolum.com/story/enkrypt-ai/ · added 2026-11-08 · updated 2026-11-08

**In short.** Enkrypt AI (Security tests, guardrails and compliance checks for enterprise AI agents and MCP servers, United States): Acquired by Anaconda with its team, August 2026. Evidence grade E: no revenue figure was checked. Team: 2 co-founders, Sahil Agarwal and Prashanth Harshangi (VentureBeat, February 2024); 16 people named on the team page in May 2026, plus interns; the team joined Anaconda in August 2026. Launched June 2022; sold 50 months after launch.

Security tests, guardrails and compliance checks for enterprise AI agents and MCP servers · United States · AI-native · started small, grew after funding · sold · group: Developer tools and coding agents

- Outcome: Acquired by Anaconda with its team, August 2026 (evidence grade of the headline number: E; no revenue figure)
- Launch: June 2022; months from launch to the outcome: 50
- Team: 2 co-founders, Sahil Agarwal and Prashanth Harshangi (VentureBeat, February 2024); 16 people named on the team page in May 2026, plus interns; the team joined Anaconda in August 2026
- Outside funding: US$2.35M seed round led by Boldcap, announced 27 February 2024 (VentureBeat and the company's release); no later round found in the sources read
- Main factor: Regulation turned into enforced rules
- Meters, 1 to 4: evidence strength 1, copyable pattern 1, capital needed 3
- Tags: AI-native, Small team, Funded, Exit, B2B, Cybersecurity

Links:

- Visit enkryptai.com: https://www.enkryptai.com/

**The product.** A security layer for companies that run AI models and agents: automated attack tests before release, guardrails that check prompts and answers at run time, a scanner and a gateway for MCP servers, and compliance reports mapped to the NIST AI Risk Management Framework and the EU AI Act. The company's pricing page lists a free plan, paid self-serve plans at US$149 and US$1,499 a month, and a custom Enterprise plan.

**Why it worked.** Large companies wanted to put language models and agents into production, but their security teams, boards and lawyers needed proof that the systems could not be tricked into harmful answers or leak data. Enkrypt sold that proof as software, and in August 2026 Anaconda, the company behind a Python and data-science platform, bought it to add a security layer to its platform, after it had bought the coding-agent company Kilo Code in July 2026.

Evidence grades: A = disclosed by a public company, a filing or a payment processor, or verified by one; B = two or more independent reports agree, or a journalist confirmed it; C = a founder claim that no one else verified; D = reported, not verified: one founder claim or a third-party estimate with no second source; E = no revenue figure found, so the headline is a non-money fact and the other facts carry their own grades; calc = computed on this page from graded facts.

## Summary

Enkrypt AI sold software that tests and guards AI systems inside companies. Before release it runs red teaming, automated attacks that try to make a model or an agent break its rules, across more than 300 attack categories. In production it applies guardrails, checks on each prompt and each answer that block a jailbreak (a prompt written to get around a model's safety rules) or a leak of sensitive data, and it scans MCP servers, programs that give an AI agent tools through the Model Context Protocol. A compliance module turns the NIST AI RMF (the AI Risk Management Framework of the US standards institute) and the EU AI Act (the European Union's law on AI systems) into rules that the software enforces.

Two Yale PhDs, Sahil Agarwal (CEO) and Prashanth Harshangi (CTO), started the company; the oldest archived copy of its site, from 9 June 2022, offered an "AI powered web3 toolkit" for personal data privacy. By February 2024 it was a Boston company selling a control layer for generative AI, and VentureBeat reported that it had raised a US$2.35 million seed round led by Boldcap and was still pre-revenue. Its archived pricing page listed paid self-serve plans in February 2026, and its team page named 16 people plus interns in May 2026. On 4 August 2026 Anaconda announced that it had acquired Enkrypt AI; the terms were not disclosed.

This page has no sale price and no revenue figure, so its grade is E. The sale is stated in an announcement signed by the buyer's chief executive and in press reports; the paid plans and the team size come from the company's own pages and archived copies of them.

## Pricing

Enkrypt AI prices in November 2026, from the company's pricing page; the same plans and prices appear on an archived copy of 14 February 2026, before the sale (grade C)

| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Explore | 0 | 0 | 500 credits to start, 50 a month; guardrails, dashboards, API, basic red-team tests |
| Launch | 134 | 149 | 5,000 credits to start, 250 a month; compliance mapping, MCP server scanning, PII masking |
| Scale | 1,349 | 1,499 | 10,000 credits to start, 1,000 a month; multimodal red teaming, MCP gateway, audit reports, SSO |
| Enterprise | custom | custom | VPC or on-premises deployment, custom service levels |

Usage is metered in credits. Each paid plan includes a starting allowance and a monthly allowance; the page does not say what one credit buys. The Enterprise plan adds role-based access, deployment in the customer's own cloud network (VPC) or on its own servers, and unlimited usage in a VPC.

## How much did Enkrypt AI make, and what was it sold for? The numbers, graded

A two-founder AI security company that raised a seed round, listed paid plans and was bought by Anaconda in August 2026; no price or revenue was disclosed

| Fact | Value | Grade | Note | Sources |
|---|---|---|---|---|
| Sale | acquired by Anaconda, announced 4 August 2026 | A | The announcement is signed by David DeSanto, CEO of Anaconda, and Sahil Agarwal, co-founder and CEO of Enkrypt AI. SecurityInfoWatch quotes DeSanto: "We're thrilled to welcome the Enkrypt AI team to Anaconda." | [1, 2] |
| Sale price | not disclosed | B | Dealroom: "Terms were not disclosed." Startup Fortune: the price "neither company has disclosed". | [3, 4] |
| Founders | 2: Sahil Agarwal (CEO) and Prashanth Harshangi (CTO), both Yale PhDs | B | VentureBeat, 27 February 2024, and the company's funding release of the same day. | [5, 6] |
| Start of the company | 2022 | C | The funding release says "Founded in 2022"; VentureBeat wrote "Founded in 2023". The oldest Internet Archive copy of enkryptai.com is from 9 June 2022 and shows "© Enkrypt AI 2022" with an "AI powered web3 toolkit" for user identity and data privacy, an earlier product. | [6, 5, 10] |
| Seed round | US$2.35 million, led by Boldcap, 27 February 2024 | C | The company's announcement, reported by VentureBeat. Berkeley SkyDeck, Kubera VC, Arka VC, Veredas Partners and Builders Fund also took part. | [5, 6] |
| Revenue stage, February 2024 | pre-revenue | B | VentureBeat: "still at the pre-revenue stage", with the product, then called Sentry, "being tested by mid to large-sized enterprises" in finance and life sciences. | [5] |
| Paid plans before the sale | US$149 and US$1,499 a month, plus a free plan and a custom Enterprise plan | C | Internet Archive copy of the pricing page, 14 February 2026. The live page in November 2026 shows the same prices. | [8, 7] |
| Team before the sale | 16 people named, plus interns, 20 May 2026 | C | Internet Archive copy of the About page: 2 co-founders, a chief product officer, a chief security officer, engineers, researchers and go-to-market staff. | [9] |
| Customer quotes on the home page | 5, of which 3 are named (Skyhigh Security, NetApp, Phot.AI) | C | Two quotes carry only a role and an employer: "Head of AI Deployment" at an "Insurance Services Company", and "AI Architect" at "Salesforce.com". Read on 8 November 2026. | [11] |
| MCP scan, June to August 2026 | 268,000 tools on 25,000 MCP servers; 143,000 vulnerabilities in 73% of the servers | C | The company's own figures in the sale announcement, with no method published. | [1, 4] |
| Model-provider partnership | OpenAI compliance integration partner; Anthropic support "coming soon" | C | Dealroom's report of the sale. The site's Compliance page offers to "Scan your OpenAI workspace". | [4, 11] |
| Time from the first archived page to the sale | 50 months | calc | Calculated: June 2022 to August 2026. | [10, 1] |

How to read this table: a grade A value is stated in the sale announcement signed by the buyer's chief executive. A grade B value is reported by a journalist. A grade C value is the company's own statement on its pages, in archived copies of them or in its press release, or a report by Dealroom. A "calc" value was computed on this page. No money amount for the sale or for revenue exists in the sources.

## When did Enkrypt AI launch, and how did it grow? Timeline

From a web3 privacy page in June 2022 to a sale to Anaconda in August 2026, in 50 months. Each entry carries its grade.

- **9 June 2022** (grade C [10]): Oldest archived copy of enkryptai.com: an "AI powered web3 toolkit" for user identity and data privacy, "© Enkrypt AI 2022".
- **27 February 2024** (grade B [5, 6]): A US$2.35 million seed round led by Boldcap, the company announced. The product, Sentry, is a control layer for generative AI, tested by enterprises; VentureBeat calls the company pre-revenue.
- **14 February 2026** (grade C [8]): The archived pricing page lists a free plan, self-serve plans at US$149 and US$1,499 a month and a custom Enterprise plan.
- **20 May 2026** (grade C [9]): The archived team page names 16 people plus interns.
- **June to August 2026** (grade C [1]): Enkrypt scans more than 268,000 agent tools on 25,000 MCP servers and reports vulnerabilities in 73% of the servers.
- **4 August 2026** (grade A [1, 2, 4]): Anaconda announces that it has acquired Enkrypt AI. Terms are not disclosed; products, plans and support continue unchanged.

## Why did Enkrypt AI work?

### Regulation turned into enforced rules

Security and compliance teams must answer to frameworks such as the NIST AI Risk Management Framework and the EU AI Act. Enkrypt sold software that maps those texts to guardrails and audit reports, so the buyer gets evidence instead of a manual audit. The acquisition post connects this to the EU AI Act obligations that applied from 2 August 2026.

### Research that shows the problem

The company published red-team reports on models such as DeepSeek-R1, a safety leaderboard of more than 200 models, and scan data on MCP servers. Each report shows a buyer a concrete risk that the product then tests or blocks.

### Self-serve plans next to enterprise sales

The pricing page lists a free plan and paid plans of US$149 and US$1,499 a month for small teams, the company says, while large companies get a custom contract with deployment in their own network.

### A place inside a model provider's product

Dealroom reports that Enkrypt was an OpenAI compliance integration partner, giving ChatGPT Enterprise customers audit and guardrail controls. That puts the product where enterprise buyers already work.

## How it was built

These details come from the company's pages, its press release and the sale announcement. They are grade C.

- Six products on the site: Agent Guardrails, Agent Red Teaming, Agent Policy Engine, AI Data Risk Audit, MCP Gateway and MCP Scanner.
- Red-team tests cover more than 300 attack categories. The research team tested models from Anthropic, Mistral, OpenAI, Gemini and DeepSeek and found attack categories that worked against each.
- Guardrails can run inside the customer's own environment; the Enterprise plan offers a private cloud network (VPC) or the customer's own servers.
- Compliance scans connect to a company's OpenAI workspace; Dealroom reports that Anthropic support was coming.
- Usage is sold in credits, with a monthly allowance in each plan.
- The 2024 product was called Sentry and was described as a control layer between language models and end users.

Sources: [1, 7, 4, 5, 11]

## Who tried this before Enkrypt AI, and why did they fail?

Three companies in our database of failed startups sold security or privacy as a separate layer and failed. The values below are from that database and are not independently verified.

The failed startups come from our database of dead startups, retold here in our own words; Its entries are AI-assisted summaries of public sources, so they carry no evidence grade.

### ThinAir (2015 to 2018; funding US$4.5M; cause: Competition)

- Promise: Real-time data-loss prevention and insider-threat detection for companies.
- Why it died: Large security vendors filled the market, and its detection produced false positives at high cost, the database says.
- Record: 

### UnifyID (2015 to 2021; funding US$20M; cause: No Market Need)

- Promise: Log in without passwords: machine learning recognizes a person by walk and typing rhythm.
- Why it died: Privacy concerns, cash-flow problems and the pandemic, the database says.
- Record: 

### Graphite Docs (2018 to 2020; funding US$400K; cause: No Market Need)

- Promise: A private, encrypted alternative to Google Docs built on a blockchain.
- Why it died: Users preferred free and fast tools to private and expensive ones, the database says.
- Record: 

## What changed

Graphite Docs and UnifyID sold privacy and security to people who were not required to buy them, and ThinAir sold security into a market full of large vendors. Enkrypt's own first page in 2022 looked like Graphite Docs: a web3 privacy toolkit for individuals. It moved to a buyer with a deadline: large companies that were deploying language models and agents and had to show their boards and regulators that the systems were safe.

Two things were different by 2024. The attack surface was new, so there were no large incumbent products for jailbreaks, agent tools or MCP servers. And a law, the EU AI Act, plus the NIST framework gave buyers a list of controls to prove. Enkrypt did not have to win the whole market alone either: it ended as a part of a larger platform, Anaconda, which the sources say wanted a security layer next to its packages and its coding agent.

## Can one person copy Enkrypt AI now?

**Verdict: The company, no. A narrow piece of it, maybe.**

Enkrypt needed a research team, a seed round, enterprise sales and 16 people before its sale, so one person cannot rebuild it. A solo founder can copy one narrow part for a smaller buyer: one check, sold self-serve, for one type of AI system that a regulation or a platform makes urgent.

1. Pick one new attack surface that buyers already worry about, for example the tools that AI agents call.
2. Publish your own scan data about it, with the method, before you sell anything.
3. Map each check to a named rule in a framework the buyer must follow, so the report doubles as audit evidence.
4. Offer a self-serve plan next to a custom contract, and list the prices.

**What is still open.** Checks for one industry's AI rules (banks, insurers, health care), security reviews for one agent platform's add-ons, and compliance reports for small companies that use one model provider's business plan.

## What could go wrong for Enkrypt AI?

- **What is not verified.** No sale price, revenue figure or count of paying customers was found. We read the joint sale announcement, Dealroom's item, SecurityInfoWatch, Startup Fortune, the 2024 seed coverage, the pricing and team pages and archived copies of them. Anaconda's own copy of the announcement (anaconda.com/blog) answered with a bot check, so it was not read. VentureBeat called the company pre-revenue in February 2024, and no later revenue statement was found. The founding year is 2022 in the company's release and 2023 in VentureBeat. Two of the five customer quotes on the home page carry only a role, not a name ("Head of AI Deployment" at an "Insurance Services Company", and "AI Architect" at "Salesforce.com"). The headcount comes from the archived team page of May 2026 only. A reader could look for a statement by the lead investor, Boldcap, or an Anaconda filing or interview that gives the price.
- **Competition and dependence on model providers.** On this site, Guardrails AI (output checks for language models, acquired by Harvey in September 2026) and PromptGuard (a firewall for AI requests) sell similar checks. Enkrypt's compliance product depends on OpenAI's enterprise workspace, as an OpenAI compliance integration partner (Dealroom), with Anthropic support announced as coming; a change in those providers' compliance interfaces would affect it. The sources do not say which models run its own guardrails.
- **The team grew after funding.** Two co-founders started the company; after a seed round in February 2024 the team page named 16 people plus interns in May 2026. The whole team joined Anaconda, the buyer's chief executive said.
- **An earlier product.** The 2022 site sold an "AI powered web3 toolkit" for personal data privacy, with no AI security product. The move to security for generative AI is visible by February 2024.

## Sources

1. Enkrypt AI blog, 4 August 2026: What It Means To Secure AI on Your Own Terms: Anaconda Acquires Enkrypt AI — https://www.enkryptai.com/blog/anaconda-acquires-enkrypt-ai (grade A). Read in full. Signed by David DeSanto, CEO of Anaconda, and Sahil Agarwal, co-founder and CEO of Enkrypt AI. Anaconda's own copy at anaconda.com answered with a bot check and was not read. Source for the sale, the product scope and the MCP scan figures.
2. SecurityInfoWatch, 5 August 2026: Anaconda Acquires Enkrypt AI to Expand AI Security and Governance Platform — https://www.securityinfowatch.com/industry-news/news/55395829/anaconda-acquires-enkrypt-ai-to-expand-ai-security-and-governance-platform (grade B). Read in full. Source for the buyer's quotes, the team joining Anaconda and the unchanged products, plans and support.
3. Startup Fortune: Anaconda buys Enkrypt AI after finding holes in most AI agent servers — https://startupfortune.com/anaconda-buys-enkrypt-ai-after-finding-holes-in-most-ai-agent-servers/ (grade B). Read in full. Source for the undisclosed price and the date of the sale.
4. Dealroom News: Anaconda buys Enkrypt AI to plug enterprise agent security gaps — https://dealroom.co/news/143042-anaconda-buys-enkrypt-ai-to-plug-enterprise-agent-security-gaps/ (grade C). Read in full (downloaded by the pipeline on 8 November 2026). Source for the undisclosed terms, the OpenAI compliance partnership and the Kilo Code acquisition in July 2026.
5. VentureBeat, 27 February 2024: Exclusive: Enkrypt raises seed round to create a 'control layer' for generative AI safety — https://venturebeat.com/ai/exclusive-enkrypt-raises-seed-round-to-create-a-control-layer-for-generative-ai-safety (grade B). Read in full. Source for the founders, the seed round, the Sentry product and the pre-revenue stage.
6. Tech Startups, 27 February 2024: Enkrypt AI secures $2.35M in funding to build visibility and security infrastructure for Generative AI — https://techstartups.com/2024/02/27/enkrypt-ai-secures-2-35m-in-funding-to-build-visibility-and-security-infrastructure-for-generative-ai/ (grade C). Read in full. Carries the company's release: founding year 2022, the founders' roles and the investors.
7. Enkrypt AI pricing page — https://www.enkryptai.com/pricing (grade C). Read in full on 8 November 2026. Source for the pricing table.
8. Enkrypt AI pricing page, Internet Archive copy of 14 February 2026 — http://web.archive.org/web/20260214102729/https://www.enkryptai.com/pricing (grade C). Read in full. Shows the same paid plans before the sale.
9. Enkrypt AI About page, Internet Archive copy of 20 May 2026 — http://web.archive.org/web/20260520030158/https://www.enkryptai.com/company/about-us (grade C). Read the team section. Source for the 16 named people and the two co-founders' titles.
10. Enkrypt AI home page, Internet Archive copy of 9 June 2022 — http://web.archive.org/web/20220609143810/https://enkryptai.com/ (grade C). Read in full. Oldest copy of the domain in the Internet Archive; source for the 2022 web3 privacy product.
11. Enkrypt AI home page — https://www.enkryptai.com/ (grade C). Read in full on 8 November 2026. Source for the product menu, the customer quotes and the OpenAI workspace scan.
12. Database of failed startups, local export — /method/#dead (grade C). Local export. AI-written summaries, not independently verified. Source for the failed predecessors.

Contact: aiolum@livyden.com. Run Enkrypt AI? Send corrections or a verified figure to aiolum@livyden.com (subject: Correction: Enkrypt AI).