Frequently Asked Questions
The questions buyers actually ask before working with us, answered the way we answer them in person: directly, including the cases where the honest answer is "you do not need us."
Consulting
Why hire an agentic AI consultant instead of having our own engineers learn LangGraph?
If your team has slack capacity and no deadline pressure, learning in-house is a fine choice, and we say so in writing during discovery. A consultant makes sense when you need production reliability on the first build, not the third: the failure modes of agentic systems (silent state corruption, validation gaps, runaway costs) are learned the expensive way. Most engagements end with your engineers owning the system, not with a dependency on us.
How long does a typical engagement take?
Discovery runs one to two weeks. A single-pipeline validation layer typically ships in two to four weeks. A multi-agent system with scheduled execution and human review gates runs six to twelve weeks depending on integration surface. We work in shippable increments, so you see working software early, not a report at the end.
We are a small team without a data platform. Are we too early?
Possibly, and that is a real answer we give. Agentic workflows need something to act on: if your data is spreadsheets and goodwill, the first engagement is data engineering, not agents. We do both, and we will not sell you a multi-agent system when a well-built pipeline and a dbt project solve the actual problem.
Does our data leave our environment during an engagement?
No. We build inside your infrastructure: your cloud accounts, your repos, your keys. Where LLM calls are involved, you choose the provider and the data-handling terms; systems can be designed for local models when data cannot leave the building at all.
What does consulting cost?
Every engagement starts with a fixed-fee paid discovery phase, which produces a written findings report and a fixed-price proposal for the build. Costs scale with scope: a single-pipeline validation layer is a different engagement than a production multi-agent system. Full details are on the pricing page.
LoreConvo & LoreDocs
Claude now has built-in memory. Why would I still need LoreConvo?
Built-in memory is a summary the vendor controls: you cannot query it, export it, version it, or carry it to another tool. LoreConvo stores full structured session records in a local SQLite database you own, searchable across projects and usable from any MCP-compatible surface (Claude, Cursor, Codex, Hermes Agent). They complement each other; they do not overlap much in practice.
What happens to my data if I cancel Pro or you shut down?
Nothing. Both products are local-first: your sessions and vaults live in SQLite files on your machine, readable with standard tools. Cancelling Pro re-applies the free-tier caps for new writes; it never locks or deletes existing data. There is no server side to shut down between you and your data.
Do LoreConvo and LoreDocs work outside Claude?
Yes. Both are standard MCP servers, verified on Claude (Code and Desktop), Cursor, Codex, and Hermes Agent. Any MCP-compatible client can use them; CLI commands cover human access without an AI in the loop.
Is my conversation data sent to your servers?
No. All storage is local. The only network call the products make is license validation for paid tiers. This is a design commitment, not a current-version accident: it is why the products exist.
How is LoreDocs different from Notion or Obsidian with a plugin?
Notion and Obsidian are human-first tools that AI can reach into. LoreDocs is AI-first: vaults are designed for context injection with token budgeting, priority tiers, and 36 MCP tools for programmatic access. If your primary reader is a human, use a docs tool. If your primary reader is an AI agent that needs the right reference at the right time, that is what LoreDocs is for.
Question not covered here? Pricing details are on the pricing page, and everything else is one message away.
Ask Us Directly