The developer’s marketing playbook
Build an AI SDR research workflow in your coding agent
Before your agent writes a sales email, ask it to explain why that company belongs on the list.
That question gives an AI sales development representative (SDR) workflow a useful starting point. Research a business, check its fit, and prepare a relevant introduction. Saka helps the agent discover businesses and inspect source evidence.
Begin with a narrow buyer description
Imagine you built a tool that helps web agencies prepare client reports. “Every agency” is a broad research brief. “Small agencies that publicly offer ongoing website maintenance and monthly reports” gives the agent evidence it can look for.
Ask it to find up to ten candidates. For each one, save the services page and the passage that suggests a fit. Do not let the agent infer that the agency is unhappy, understaffed, or shopping for software unless a source supports it.
The difference matters in the resulting email. “Your team must be wasting hours” asserts a problem you have not verified. “Your services page includes monthly reporting” is an observation the recipient can recognize.
Keep four decisions separate
First, does the company plausibly benefit from the product? Second, who would care about that benefit? Third, do you have an accurate business contact? Fourth, is the planned outreach appropriate for that recipient and permitted by the sending service?
A search result only begins this work. It does not verify an address or enroll someone in a campaign. Record uncertain fields as unknown. A shorter list with evidence is easier to review than a large list of guesses.
The research prompt saves a CSV and account briefs with source URLs, access dates, observations, and uncertainty.
Make the first message useful
For the fictional reporting tool, offer a sample using dummy data. That gives the recipient a concrete way to evaluate relevance before committing to a meeting.
An illustrative introduction might be:
Subject: client report sample
Your maintenance page includes monthly client reports. I’m building a tool that turns website metrics into a report clients can read. Would a sample using dummy data be useful to compare with your current report?
Alex
Use this only when the observation and capability are true. Replace the fictional sender, resolve the contact details, and include the required sender and opt-out information before delivery.
Follow-ups should add something useful, such as a specific example or an answer to a likely question. Repeating the same ask provides little reason to respond.
Decide what happens after a reply
A sequence can become awkward quickly if it continues after someone answers. Any reply should pause follow-ups. The owner can then review a suggested response.
An opt-out suppresses future outreach. A bounce or complaint should also stop delivery. An uncertain reply belongs in a review queue. These rules need to work in the sending integration, including when events arrive twice or a worker restarts.
Saka does not provide that sending system. It handles research. Your agent can prepare the integration, while an email provider and a persistent workflow handle delivery and events.
Start with a review queue
Your first milestone is a set of sourced account briefs and messages worth reviewing. Then follow the launch guide to connect delivery.
Measure qualified conversations and read the actual replies. If nobody recognizes the problem, investigate the targeting and offer before expanding the list. Use the weekly review prompt to record one testable change.
You can begin research without buying a contact database. That does not make the complete sales workflow free: agent usage, contact verification, mailboxes, and sending may cost money. Keep those decisions visible as the workflow grows.
Keep the workflow in your coding agent
Install the free ai-sdr skill with npx skills add email-marketing-skills/email-marketing-skills --skill ai-sdr. It works from your project context and available tools. Saka business research is optional; live delivery uses your chosen email provider. Follow the quickstart for setup and the plain-text fallback.