Welcome to this month’s edition of boberdoo Lab, our monthly dispatch where we share early looks at upcoming tools we are building behind the scenes.
This month, we are diving into our latest project: bringing an MCP server to the boberdoo lead distribution platform. Instead of jumping between admin screens, manual setup forms and multi-tab diagnostics, you will soon be able to run core operational workflows simply by asking an AI assistant to handle the job.
The boberdoo MCP Server
Model Context Protocol (MCP) is an open standard that allows AI assistants like Claude to interact directly with external software through structured, strictly typed tools.
Instead of training a separate chatbot or relying on brittle UI automation, an MCP server gives your AI assistant, like Claude or ChatGPT, a direct tool catalog. Each tool performs one specific action in your boberdoo system and executes with the same validation logic and business rules as our admin interface. The assistant inspects each tool's required parameters, asks for any missing inputs, executes the request and verifies the platform’s response.
Our initial release delivers 19 tools across four key operational areas:
- Partner Onboarding and Management
- Filter Sets (Pricing, Targeting and Delivery)
- Lead Capture Forms
- Performance and Troubleshooting
1. Partner Onboarding and Management
Setting up a new buyer typically requires filling out an account profile, sending an activation email and manually building their filter sets from scratch. Our onboarding tools consolidate that entire sequence into a single natural-language request.
signup-partner: Runs the entire signup wizard in a single call. It provisions the buyer account, duplicates filter sets from global templates or an existing partner, and retargets each copy by state, custom ZIP list or ZIP radius.
create-partner: Creates a partner account, returns the activation link and automatically sends the account creation and password setup email.
list-partners: Searches partner accounts by company name or primary contact, with filtering by status.
get-partner-info: Fetches complete contact details, mailing addresses and profile metadata for an individual partner.
set-partner-status: Changes an account's operational state to Active, Temporarily Stopped or Not Active, with an optional log reason.
Lab Use Cases
- Onboarding lookalike buyers: “Sign up Roofing Company and give them the same filter sets as our top roofing buyer, but within 40 miles of 60614.” The assistant provisions the account and builds the retargeted filter sets in one step.
- Immediate account pausing: “Temporarily stop Mortgage Company; they’re over their credit limit.” The assistant confirms the partner name and ID before updating the status.
- Instant account inquiries: “Who is the contact at Solar Company, and what state are they in?” The assistant pulls the partner's profile and immediately returns their contact and address details.
2. Filter Sets: Pricing, Targeting and Delivery
Filter sets dictate how leads are matched, what buyers pay and how data routes downstream. These tools let the assistant build routing logic from plain-English descriptions and verify exactly what was saved.
get-filter-set-fields: Inspects a lead type's schema and lists every parameter it accepts, including required fields, accepted criteria values, state options, ZIP targeting modes and delivery types.
create-filter-set: Builds a brand-new filter set complete with bid pricing, allowed sources, match priority, volume limits, delivery endpoints, schedules and criteria rules.
update-filter-set: Applies targeted updates strictly to the fields you specify while keeping the rest of the filter set unchanged.
get-filter-set: Retrieves and displays an individual filter set in full detail, matching the administrative view.
get-filter-sets: Lists all filter sets assigned to a partner, displaying their status, priority, state coverage and allowed traffic sources.
set-filter-set-status: Toggles an individual filter set on or off.
list-filter-set-templates: Browses your library of global filter set templates organized by lead vertical.
copy-filter-set: Clones a single filter set template or another partner’s set directly to a buyer, with optional geographic overrides.
copy-partner-filter-sets: Clones every filter set from one partner to another. New copies are set to inactive by default so they can be verified before going live.
Lab Use Cases
- Conversational creation: “Create a $25 filter for Pond Management leads, ZIP 46360 only, New Pond projects only, no delivery.” The assistant checks accepted field values, creates the record, and reads back the saved configuration.
- Surgical repricing and retargeting: “Raise Auto Company’s Illinois filter to $32 and add Indiana.” Only those target parameters are adjusted.
- Territory scaling: “Copy our standard HVAC template to Heating Company for a 25-mile radius around 46360, same state only.”
- Pre-change audits: “Show me every active filter set for Insurance Company and what states they cover.”
3. Lead Capture Forms
The create-form tool works like the Forms wizard, letting you generate live, tracked lead-capture forms on demand.
It returns the direct embed URL, copy-paste website embed code and a configuration link for further field adjustments. You can specify lead sources, landing page URLs, Pub ID and Sub ID tracking variables, custom or built-in stylesheets and post-submission actions, such as a tokenized thank-you message (e.g., using {First_Name}) or a URL redirect.
Lab Use Cases
- Instant landing page deployment: “Make a Pond Management form for the ‘pond’ source on our new dredging page, with our stylesheet, redirecting to /thanks.” The assistant instantly produces ready-to-paste embed code.
- Granular campaign attribution: Rapidly spin up unique form variants with distinct Sub IDs for individual ad sets to keep source reporting clean.
4. Performance and Troubleshooting
When a lead doesn’t sell, diagnosing why usually means comparing the lead’s attributes against dozens of buyer requirements. These tools run root-cause diagnostics and provide quick visibility into performance.
get-unmatched-reasons: Re-runs match logic for a specific lead as a dry-run simulation without delivering data, storing records or charging buyers. It breaks down why each partner or filter set passed, categorizing results by rejection reason, filter rule and failing attribute. It can also simulate routing as if all partners and filter sets were active.
get-partner-dashboard: Summarizes lead volumes and sales figures per partner for today, yesterday, the last 7 days and the prior 7 days, plus cumulative totals.
get-source-dashboard: Displays real-time vendor activity for any lead type, detailing incoming and matched pings, posts, rejection counts, bid/buy rates and total sales.
list-lead-types: Lists active lead types across your system so the AI assistant always references valid vertical IDs.
Lab Use Cases
- Failure diagnosis: “Why didn’t lead 48213 sell?” The assistant can immediately surface that 12 buyers excluded the lead’s ZIP code, 4 hit their daily budget cap and 2 didn't accept the traffic source.
- Pre-launch simulation: Re-run matching on historical leads as if a newly onboarded buyer were active to verify their criteria before routing live traffic.
- Morning performance checks: “Which partners are down the most versus last week, and which sources have the worst buy rate today?”
- Early vendor anomaly detection: Sort inbound sources by rejected posts to catch integration errors or malformed payloads before they affect daily margins.
Connecting the Tools: Chained Workflows
The primary benefit of the MCP architecture is that an assistant can chain multiple discrete tools together in a single conversation:
- Workflow 1: Onboarding a buyer in minutes
A rep tells the assistant: “Sign up Ponds Company. $25 per New Pond lead in ZIP 46360, no delivery to start.” The assistant creates the partner profile, references the Pond Management field parameters, builds the filter set and reads back the full configuration for review. - Workflow 2: Isolating routing bottlenecks
When the source dashboard shows incoming traffic matching poorly, you can run get-unmatched-reasons across recent leads to isolate whether buyers are hitting volume caps or excluding specific ZIP codes. The assistant identifies the bottlenecked filter sets and modifies them upon your approval. - Workflow 3: Scaling an ad campaign
Spin up a tracked form variant for a landing page, duplicate your primary filter set template across three expanding buyers and check the partner dashboard the next day to monitor lead acceptance.
Architectural Guardrails and Safety
Because these tools interact with routing logic and data, safety controls are built into every level of the integration:
- Explicit Confirmation: Changing operations require explicit user confirmation before changes are applied.
- No Unchecked Bulk Changes: When an action touches multiple records, the assistant lists the affected items, waits for explicit approval, processes records sequentially and halts immediately on an error.
- Safe Defaults: Cloned filter sets start in an inactive state for manual review. The unmatched diagnostics tool runs strictly as a dry-run simulation without triggering deliveries or billing transactions.
- Read-Back Verification: The assistant reads saved filter set states back from the system so operators verify actual database values, not just what was requested.
- Isolated Billing: Partner provisioning tools do not configure credit cards or payment processing.
- Prompt Injection Protection: External partner names, source tags and form fields are processed strictly as data values and are never parsed as instructions by the assistant.
What’s Next for the Lab
The boberdoo MCP tools are currently running in pilot testing on our internal systems. Because MCP is an open standard, the server will work with any MCP-compatible environment, meaning your team can leverage the AI assistants you already use.
While this initial release centers on partner management, filter set configuration, form building and diagnostics, we are actively scoping future capabilities. If you would like early access to the pilot or want to share feedback on which administrative tasks you would like to automate next, reach out to your boberdoo account manager or request a demo to see it live.
