Vellum
Always-on personal AI assistant — the LLMOps platform pivoted in 2026
Overview
Vellum is an always-on personal AI assistant with a three-layer memory system: workspace files, a curated knowledge base, and auto-extracted long-term memory with confidence scoring. It handles calendars, email, files, browsing and messaging, runs on the company's managed platform or self-hosted, and is open source. The company sold an LLM application development platform — prompt management, evaluations, workflows, observability — until it pivoted in 2026; those product pages now redirect to the homepage.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 12 Sep 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- First tasks included
- Small computer, 1 vCPU / 2 GiB
- 10 GB storage
- No base fee
- Medium computer, 2.5 vCPU / 5 GiB
- 30 GB storage
- Includes the $10/mo base fee
- Large computer, 4 vCPU / 8 GiB
- 60 GB storage
- Includes the $10/mo base fee
- No platform fee
- No computer tier charges
- You hold the keys and the uptime SLA
Use cases
What you can produce with Vellum
- Run an always-on assistant that handles calendar, email, files and browsing
- Give it a three-layer memory — workspace files, a curated knowledge base, and auto-extracted long-term memory
- Reach it from desktop, mobile, web or messaging rather than one app
- Self-host it open source on your own hardware, with no platform fee and your own keys
- Budget in credits, where $1 = 1 credit and you pay for what the assistant actually does
- Note the $10/mo base fee on Super and Ultra, which Mighty does not carry
Vellum is an LLM application development platform combining a visual workflow editor, prompt management/versioning, systematic evaluation and deployment without code redeploys. It is model-agnostic across OpenAI, Anthropic, Google and Cohere, carries SOC 2 Type II and HIPAA compliance, and its evaluation suite is a genuine strength for teams that need to measure and regression-test output quality rather than ship vibes.
It suits cross-functional product teams shipping LLM features who want collaboration between engineers and non-engineers, plus rigorous evals and staging-to-production testing. Caveats: it competes in a crowded, fast-moving space against open frameworks and rival platforms, usage-based pricing can climb with volume, and committed adopters take on platform lock-in. The API and SDK support are solid. Globally available and English-centric, so usable across ASEAN with no region-specific provisions.
What people say
Vellum has grown from a prompt-management niche into a fairly complete LLM-operations platform, and in January 2026 it pushed further with "Vellum for Agents", which assembles agent workflows from a natural-language brief. The company remains independent, is SOC 2 Type II and HIPAA compliant, and offers a free tier alongside hosted and private-deployment options — attributes that matter to the regulated product teams it courts.
User sentiment on review platforms is strongly positive. Reviewers on G2 and Capterra (where it surfaced at around 4.8/5) repeatedly single out two things: prompt versioning with side-by-side model comparison, and customer support that is described as unusually hands-on — teams say Vellum staff help them ship rather than pointing at docs. Teams that previously managed prompts in spreadsheets describe the move as transformative, with several reporting AI feature cycles shrinking from weeks to days because product managers and engineers can iterate in parallel instead of queuing behind one another.
The criticism is more structural than functional. The most common frustration is the seat cap on standard plans — growing teams get pushed into enterprise sales conversations sooner than expected, and at least one independent 2026 review scored Vellum modestly on the grounds that pricing, features and ease of use do not balance well for non-technical teams. Evaluation setup takes real engineering effort to get right, and reviewers note that Vellum's eval suite, while well integrated with its workflows, lags behind dedicated evaluation products on depth.
Vellum fits product teams shipping LLM features to production who want prompts, tests, workflows and monitoring in one governed place — especially in healthcare or other compliance-heavy settings. Solo developers and teams that only need lightweight prompt experiments will find cheaper, simpler options, and anyone expecting a no-code experience should budget for engineering involvement.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Vellum's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Vellum unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Vellum directly →
Spotted something out of date? Suggest an update →
Alternatives to Vellum
More in Productivity
