This guide is for HR and recruiting teams rolling out Tactiq, with Claude as an optional extra. It covers what to set up, in what order, and how to know it is working. For the prompts and day-to-day habits, see Tactiq + Claude for recruiting: best practices.

Every interview leaves a scorecard, every scorecard sits in its role's Space, and each hiring round sharpens the criteria. Tactiq creates the scorecards.

Every team starts at Stage 1. If someone on your team already has a scorecard prompt that works, use it as the kit in Stage 2 rather than writing a new one.

One Tactiq Space per open role, and every interview named the same way: role, round, candidate. This is what tells AI what each meeting actually is. Without it, AI sees "a meeting at 10am." With it, AI sees "second-round interview, Senior Backend, candidate Y."

Done when: every interview from the past week sits in the right role Space, labelled.

A shared Tactiq kit produces a scorecard after each interview, ideally set up as a workflow so it runs on its own. Each one holds four things:

The rules for fair assessment are written into the Tactiq kit's prompt, so they apply every time it runs. Nothing about age, accent, appearance or other protected traits, and every judgment must point to something the candidate said. Use the same rules whenever you compare candidates later, so everyone is judged the same way.

Starting from scratch? The scorecard prompt in Tactiq + Claude for recruiting: best practices is ready to copy into your kit, with these rules already built in.

Setting up the workflow. Once the role Space exists, the simplest route is a workflow set up inside the Space, which suits one recruiter's interviews. For the whole team, use the team workflow setup. Our customer success and support team can help with either.

Done when: every interview that week has a scorecard.

Compare candidates from their scorecards. Check the transcript before a decision is made.

Why scorecards come first. The kit reads one interview at a time against the same criteria and rules, so every candidate lands in the same format. That's what makes candidates comparable across a round.

Tactiq for speed, Claude for follow-up. Tactiq is fastest when you know the question: run a saved prompt, or use Ask AI across the role's Space with every interview in the round selected. Claude suits open follow-up questions and digging into what the first answer turned up, and it can bring in other tools such as your ATS.

Watch for drift in follow-ups. Open back-and-forth is where answers start leaning on impressions instead of evidence, until you're back at "who's the best fit?" That question tends to reward whoever sounded most fluent, which is unfair to nervous candidates and anyone interviewing in a second language. Bring each follow-up back to the criteria and the quote behind them.

When several candidates stand out. A scorecard rates each person against the criteria on their own, so four strong candidates can all score well. To tell them apart, ask where they actually differ, with quotes. The gap is often in what was "Not covered" for some but not others, and that becomes the question for the next round.

Why transcripts still matter. A scorecard is a summary, so it can't hold every detail. "Would relocate" and "would consider relocating" can both end up as a yes. Some questions also need the full transcript: where two interviewers disagree, how consistently you interview, and which questions never got asked.

Check the quote before it counts. Before a scorecard feeds a hiring decision or a rejection, search the transcript for the quote behind the rating and confirm the wording. This is why the kit puts the candidate's exact words under every rating. It takes a minute and keeps every decision easy to explain.

Your admin connects Tactiq to Claude. Share the Tactiq Trust Center and the MCP setup guide with your IT team so they can review it first.

Use Claude for open follow-up questions across a round, or to bring in other tools such as your ATS. Give it the scorecards from Tactiq rather than full transcripts. The shorter and more consistent the input, the better the answers.

The best practices guide has ready-made Claude prompts for comparing a shortlist, prepping the next round, and working through panel disagreements.

Done when: one recruiter has run a comparison across a round in Claude.

Tactiq holds the memory and the everyday answers. Each role's Space holds every interview and scorecard, the shared kit adds to it after every call, and Ask AI answers questions across the round in seconds. Share the kit and your round questions as team prompts, so everyone works from the same memory.

Claude adds open exploration. Connected to Tactiq, it can explore a round through open follow-up questions and bring in context from your other tools. Save the questions your team asks often as Claude skills, so everyone asks them the same way.

One person on your team owns the criteria for each role, so the kit, prompts, and skills stay in step.

This is where one person's good prompt stops being theirs and becomes the team's, and every new interview makes the whole team smarter.

Done when: anyone on the team can ask a question about any open role and get an answer grounded in the team's own interviews.

Hire, or no-hire outcomes go back into the role's Space. Once a month, use Ask AI across the Space, or Claude, to compare scorecards against decisions and flag criteria that didn't predict anything. The hiring manager approves any changes to the kit, prompts, or skills.

This doesn't re-score anyone. Scorecards stay as they were made after each interview. What changes is the kit for the next role: if a criterion rated every candidate "Strong," it didn't help tell them apart, so it gets sharpened.

Done when: the kit, prompts or skills have been updated at least once based on real outcomes.

Four signs the kit is doing its job:

Need help? Contact Support