Automating Recruitment Interview Summaries with AI

Process Automation: Interview Summaries with AI

At Oktana, we constantly look for ways to improve our own productivity through automation. Somewhat unique to our industry, we have an in-house team of technical recruiters. This team conducts numerous candidate screening calls daily with the challenge of capturing and standardizing interview summaries. 

Until recently, notes were either taken entirely manually or were generated through Gemini’s “Taking Notes” feature in Google Meet. However, neither approach delivered the structured summary format the team needed for consistency when entering their notes in Salesforce. Therefore, to address this challenge, we implemented a tailored automation solution combining Google Apps Script, AI, and Salesforce. To address this, we implemented a tailored automation solution combining Google Apps Script, AI, and Salesforce.

Challenge

While recruiters had the option of using Gemini’s “Taking Notes” feature in Google Meet, the summaries generated were limited to Gemini’s default structure. As a result, this conflicted with Oktana’s preferred format, which follows clearly defined topics to ensure complete and consistent screenings and notes throughout the recruitment process. In addition, having to manually enter summaries into Salesforce introduced an operational risk, since any delays or errors could reduce visibility across the team and ultimately create a poor candidate experience.

The challenge was to: 

📝 Create a process that reliably captured interview content

📄 Generate summaries in the desired format

☁️ Seamlessly store them in Salesforce without requiring manual effort

✍️ Still provide the ability to manually enhance the notes with additional observations

Solution

We implemented an integrated automation leveraging Google Apps Script, AI, and Salesforce.

AI-Powered Transcription
Using Google Meet’s transcription capabilities, interviews are automatically captured and stored.

Google Apps Script Automation
A custom Google Apps Script processes the transcriptions and sends them to a backend service for summarization.

Python AI Service
Our backend service, built in Python, performs inference with our chosen AI model to generate a structured summary aligned with the Recruitment team’s format and requirements.

Salesforce Integration
The finalized summaries are automatically stored in Salesforce, ensuring data consistency, easy accessibility, and full visibility without manual entry.

Technologies

  1. FastAPI → Python framework for building high-performance web services and APIs.
  2. Uvicorn → ASGI server for running our FastAPI APIs.
  3. Hugging Face → Language model platform and integration library.
  4. Meta-Llama-3-70B-Instruct → General-purpose LLM for reasoning, summarization, and language analysis.
  5. Transformers → Hugging Face library for working with natural language processing (NLP) models.
  6. Novita → Compute infrastructure optimized to run AI models with greater speed and efficiency.
  7. Google Apps Script → Automates transcription processing, sends the content to the model for summarization, and stores the results in Salesforce.
  8. Gemini (Google) → Used for automatic note-taking and transcriptions in Google Meet.
  9. Salesforce (REST Apex endpoint) → Stores generated summaries in Salesforce records (Contact Tasks for Recruiting, Feedback for HR, etc.).
  10. Linode Server → Hosting for our Python backend.
  11. Chrome Extension (JavaScript) → Automatically activates the transcription feature during Google Meet calls.

Results

Oktana’s in-house technical recruiters can now focus fully on conversations with candidates, confident summaries will be generated in the expected structure and instantly available in Salesforce.

The solution eliminates the risk of inconsistent or incomplete summaries, reduces manual workload, and ensures leadership has reliable data for decision-making. 

Overall, this project clearly demonstrates how Oktana applies AI and automation internally to boost productivity while also enhancing the candidate experience. Moreover, if you’re looking for an AI-driven solution to automate manual processes, improve efficiency, and ensure consistent data capture, contact us today to explore how we can support your team.

Iris
Software Engineer

You might also like

By continuing to use this site, you agree to our cookie policy and privacy policy.