This is a guide to connecting Harmonic to your own systems and workflows. It's meant to help you think through your build, not to prescribe one path per use case.
What's available
Console: Harmonic web app
API: programmatic access via REST or GraphQL
MCP: connects Harmonic to Claude, ChatGPT, or any MCP-compatible agent
Scout: Harmonic's AI agent, reachable across Console, Slack, MCP (coming soon), and Scout API
Bulk data: the full dataset delivered to your warehouse (Snowflake, BigQuery, S3)
Chrome extension: Harmonic data on any webpage
How to think about your build
A few questions worth asking, not a checklist to answer in order:
Is this something you or your team will use live, or something meant to run without anyone using or watching it?
Are you building a one-off tool, or infrastructure other systems will sit on top of?
Do you need Harmonic's reasoning, or just its data?
Is this living inside Harmonic, inside another tool you already use, or inside something you're building yourself?
There's no single mapping from these answers to a product. The right combination depends on your stack and your team's technical depth as much as on the use case itself.
API vs. MCP: what's the difference?
Both give you access to the same underlying Harmonic data. So which should you use?
API: built for developers writing code
If you're building something and need to fetch companies, search results, update a list, or push data into another system, you write code that calls a specific Harmonic endpoint. This is the most direct, most flexible path, and the one that requires the most building. No-code tools like Zapier, Make.com, and n8n simplify this through native connectors and are great for multi-tool automations. Harmonic is also a partner of Polytomic, purpose-built for large-scale, recurring ETL into your CRM, data warehouse, and more.
MCP: built for AI tools and agents
MCP is a standard way for an AI client to discover and use Harmonic's tools without you writing custom integration code for each one. Connect once, and the AI can search, enrich, and manage lists on its own. MCP doesn't replace the API, Harmonic's MCP server runs on the same underlying platform. The difference is who's on the other end.
Where Scout fits
Scout's reasoning is reachable through either path: in Console or Slack, and coming soon through MCP, for asking questions live; or through Scout API when you're building your own system or agents and want that reasoning as a component inside it.
Scout: Harmonic's AI agent
Scout goes well beyond open-ended Q&A. It's Harmonic's AI agent, with access to the full Harmonic dataset, real-time web search, and Harmonic's own core capabilities: it can search across the full dataset, read and write to lists, and set custom fields, the same underlying tools that power Console itself.
From there, Scout can do things like deep research on a company, person, or investor, generate a full market map, draft a memo or brief, or write its own reasoning directly into a score or custom field, just a few examples.
The features that make this work:
Scheduled Scouts: set a thread to run daily or weekly, checking for new results on its own and only surfacing something when there's actually something new
Personalization: Scout remembers your fund's thesis, sectors, and preferred output format across sessions
Skills: a team's scoring rubric, memo format, or outreach style, saved once and applied automatically whenever anyone asks for that kind of output
Threads: an ongoing conversation that keeps context as you refine a search or dig deeper on a result
Scout across surfaces
Scout is the same engine wherever it shows up, Console, Slack, MCP (coming soon), or Scout API, so most of what it can do (research, memos, market maps, scoring) isn't tied to one surface. What varies is who or what can write back into Harmonic: a person approves each write in Console; a Scheduled Scout writes autonomously within the scope it was set up with; Scout API can't write to Harmonic at all; and Scout in MCP doesn't need to, since the MCP client can use its own list and field tools to act on what Scout finds.
Bulk data
The full Harmonic dataset delivered directly to your own warehouse, Snowflake, BigQuery, or S3, refreshed weekly. Teams building on bulk typically train proprietary scoring models, join Harmonic data with their own datasets for analysis, or build custom applications on top of the full dataset.
NGP Capital ($1.6B AUM) built an internal platform called "Q" on weekly Harmonic refreshes. "We use Harmonic across the entire funnel," says Atte Honkasalo, VP of Data & Analytics, "from discovering companies, to evaluating them, to understanding momentum." Their team trains scoring models directly on the structured data instead of building that data infrastructure themselves.
What customers are building
Draper ($2.3B AUM, 400+ portfolio companies) had its MBA interns build two production tools the whole investment team now uses: Draper Brain, an MCP server pulling in 15,000+ contacts and 450+ portfolio companies alongside Harmonic's live company data, and the Draper Deal Engine, which scores incoming companies across eight dimensions using the API and four decades of Draper's own investment history. "With Harmonic powering Draper Brain," says Patrick Blumenthal, who built it, "mapping a competitive landscape that used to take half a day takes minutes." Every company the Deal Engine flags also gets run through Scout as a second opinion before it reaches a partner.
Rillet, an AI-native ERP backed by Sequoia, a16z, and ICONIQ, sells through the cap table: their team runs Harmonic's MCP from inside their own agent setup to pull funds matching their ICP, then the portfolio companies, then the right contacts at each. "The MCP is what makes it click," says Tyler Robinson, who leads private-equity partnerships at Rillet. They also point Scout at an entire fund's portfolio to score it against their ICP in one pass, and use the Chrome extension to pull contact details straight off LinkedIn.
Quantum Talent, an executive search firm, uses Scout and Network Mapping to source candidates and track talent flow, seeing who's leaving which companies and why, and mapping warm paths into founders through their own team's connections. "It's a powerful way to translate years of relationship-building into new search intelligence," says Chad Rogers, Director at Quantum Talent.
Harmonic for your team
More ways teams are building with Harmonic today:
Investors and VCs
Turn a saved thesis into a running shortlist: Scout does first-pass screening on new opportunities, stack-ranks them, and hands you a prioritized list instead of a raw feed to sort through yourself
Revisit companies you passed on and get flagged automatically if something's changed since
Track talent flow at strong teams to catch spinouts and future founders before they're visible anywhere else
GTM and sales teams
Map your own product's power users to companies, then flag anyone who left for a new company as a fresh prospect, combining your own product data with Harmonic instead of treating them as separate systems
Get flagged automatically when an account spikes in engineering headcount or just raised, before it shows up anywhere else
Find warm paths into a prospect through your team's existing network before the first outreach
Talent teams
Company-first sourcing: describe the kind of team you want to hire from, and Scout surfaces the right companies first, then the right people at them, a workflow that doesn't require you to already know who to target
Track hiring and firing trends across an entire market, department by department, to catch industry-wide shifts rather than one company at a time
Watch for departures at strong teams to catch people moving into stealth or new ventures early, before they're recruiting-visible
Corporate development
Run acquisition and acqui-hire pipelines side by side, evaluating team pedigree alongside company fit
Support a portfolio company's own biz dev: when they land a new type of customer, find similar buyers and flag the ones your team already has a warm path into
Set up scheduled monitoring for new targets in your space so nothing slips by unreviewed
One last thing
Most of the setups we see aren't single-surface, e.g. a firm might use Scheduled Scouts to monitor for new opportunities, the API to keep that data synced into their CRM, and MCP so anyone on the team can ask a follow-up question without waiting on an engineer. There's no wrong order to pick these up in, start with whichever piece solves what's in front of you right now, and add the rest as the need shows up.
Get started
Reach out to your Harmonic Account Manager or support@harmonic.ai and we'll help scope your build.