CodaAI TurboInsights

A struggling team lacking resources and expertise. Witness how they save up to 30-50% of time, unlock accurate results, and build trust within their company by democratizing research.

Comment

Inspiration

I like good processes. And make sense of the mess. And do it together with the team. Unfortunately, some products and teams can afford a good research process. This inspired me to create CodaIA Turbo Insight.

What it does

It helps to go through the research process 30-50% faster without losing the quality of research insights. It also helps to democratize research by inviting non-researchers to the research process.

How I built it

I already had 3 cases with complex Coda templates for research. I empowered every step of the research process with AI.

  1. Research planning Coda + Grain + Zappier _I created 3 types of research codes in Coda codes#1, codes#2, codes#3. (Code#1 help to mark transcription chunks in Grain. Codes#1, codes#2, codes#3 are used for the automatic generation of prompts to OpenAI, supporting Coda AI autofill function and supporting widgets helping to synthesize faster. ) Then I Duplicated codes#1 in Grain. And created integration of Grain and Coda in Zapier _

  2. Interview session. Transcribing and linking transcription chunks with codes#1 is happening in a parallel Coda + Grain + Zappier __Transcription chunks linked to codes#1 together with embed video excerpts are coming from Grain automatically. (the trigger is the researcher pushing the code#1 in a meeting popup and then the button "Create transcription chunk").

  3. Organising data: withdrawing research answers from the transcription text Coda + OpenAI Code# is a research target. Each research target is linked to the research question. For each transcription chunk, OpenAI is withdrawing answers to research questions of the code#1 research target that is linked to this transcription chunk. The research answer is an Answer Title and Quote from the text.

  4. Organising data: coding research answers with codes#2 codes3# Coda + CodaAI _The usual Coda button is creating research answers in a separate base. Coda AI is linking them to codes#2, codes#3. These codes could be anything relevant to your research model: codes#2 - types of problems and codes#3 steps for CJM, codes#2 - types of research answers for affinity mapping and codes#3 - roles.

  5. Synthesising data: Coda + CodaAI Widgets with filters be codes#1, codes#2, codes#3 are helping to slice and dice data. Researchers are working on the widget "board view" and grouping similar answers to insights. CodaAI is generating a summary based on answers linked to insight. Since the answers contain relevant quotes, summaries appear to be well-rooted in the initial words of the respondent.

Challenges I ran into

The biggest challenge was to give up the idea that template will do everything by itself. The researcher's eye still seems to be crucial. I hope that the process I came up with is a golden middle between dying in endless transcriptions, data points, stickies, and too shallow analysis. The second challenge is a prompt to Open AI. I could just leave the prompt "withdraw meaningful insight". So after several iterations, I added relevant research questions (via formula) to prompt to guide AI's answers. + CodaAI autofill is amazing, thank you, guys!

Accomplishments that I'm proud of

  • Ability to save from 30% till 50% of time
  • Hybrid teams are learning through doing (and it's really cool that you can out the quality of the process into Coda's workflow, so other people just follow it and produce strong research results)
  • More accurate research results, more time for thinking though them
  • Scale the impact via democratization of research (it's a trend link

What I learned

  • how to give tasks to AI
  • how to scale research solutions for teams
  • how to plan research in a way that could be automated

What's next for CodaAI TurboInsights

For this template:

  • polish instructions (to be honest, more like make good instructions) and hints
  • more video-instructions

CodaAI TurboInsights 2.0

  • add domain-knowledge hub (I have template with AI for bases of knowledge, can adjust it to domain knowledge)

Less generic templates:

  • template for CJM
  • template for personas

More complex templates

  • Human-Centered Product Strategy Ecosystem
  • Templates for workshops facilitation

Link to PDF https://drive.google.com/file/d/1T5NaEHEfob9xQfN1fSe_7tP-vgeRE-dt/view?usp=sharing https://ooo.mmhmm.app/z_JYhd4t2yjYK6oN8CeVpr

Built With

  • coding-approach-in-qualitative-research
  • grain
  • openai
  • zapier
Share this project:

Updates